Normal CXR Report: Documenting Chest X-rays in COVID-19 Patients

28.08.197817.11.2021 by alexxlab

How does a normal chest X-ray report look for COVID-19 patients. What are the common findings in chest X-rays of ambulatory COVID-19 patients. How reliable are chest X-rays in diagnosing COVID-19 in urgent care settings.

Содержание

Understanding Chest X-ray Findings in COVID-19 Patients

The COVID-19 pandemic has significantly impacted healthcare systems worldwide, prompting urgent care centers to adapt their diagnostic approaches. One such adaptation involves the use of chest X-rays (CXRs) in evaluating patients with suspected COVID-19. A recent study conducted in the greater New York City area has shed light on the prevalence and characteristics of CXR findings in ambulatory COVID-19 patients presenting to urgent care centers.

Study Overview

The study, led by Michael B. Weinstock, MD, and colleagues, analyzed CXRs from 636 patients with confirmed COVID-19 who visited urgent care centers between March 9 and March 24, 2020. Eleven board-certified radiologists independently reviewed the CXRs, classifying them as normal, mild, moderate, or severe, and characterizing specific findings.

Key Findings

58.3% of CXRs were normal
41.7% showed abnormalities
89% of CXRs were either normal or only mildly abnormal
Most common abnormal findings: interstitial changes (23.7%) and ground glass opacities (18.9%)
Abnormalities were predominantly located in the lower lobes (33.8%), bilateral (20.9%), and multifocal (24.2%)
Pleural effusions and lymphadenopathy were uncommon

The Role of Chest X-rays in COVID-19 Diagnosis

Do chest X-rays provide reliable diagnostic information for COVID-19 patients in urgent care settings. While CXRs are widely available and relatively inexpensive, their sensitivity for detecting COVID-19-related lung abnormalities is limited. In this study, over half of the confirmed COVID-19 patients had normal CXRs, highlighting the potential for false negatives when relying solely on this imaging modality.

Why are chest X-rays less sensitive than CT scans for COVID-19 diagnosis. CT scans offer higher resolution and can detect subtle lung changes that may not be visible on plain radiographs. Studies have shown that CT scans have a sensitivity of 56%-91% for COVID-19 lung pathology, depending on the time between symptom onset and imaging. However, CT scans are more expensive and less readily available in urgent care settings, making CXRs a more practical option for initial screening.

Interpreting Chest X-ray Findings in COVID-19 Patients

How should healthcare providers interpret CXR findings in patients with suspected COVID-19. It’s crucial to understand that a normal CXR does not rule out COVID-19 infection. In fact, this study demonstrates that nearly 60% of confirmed COVID-19 patients had normal CXRs. When abnormalities are present, they typically manifest as interstitial changes or ground glass opacities, often in the lower lobes and with a bilateral or multifocal distribution.

Common Radiographic Patterns

Interstitial changes
Ground glass opacities (GGO)
Lower lobe predominance
Bilateral involvement
Multifocal distribution

Can chest X-rays help assess the severity of COVID-19. While CXRs may not be highly sensitive for diagnosing COVID-19, they can provide valuable information about disease severity. In this study, radiologists classified abnormal CXRs as mild, moderate, or severe. Of the 265 abnormal cases, 195 demonstrated mild disease, 65 showed moderate disease, and only five exhibited severe disease.

Implications for Urgent Care Practice

How should urgent care providers utilize chest X-rays in managing COVID-19 patients. Given the high proportion of normal or mildly abnormal CXRs in symptomatic COVID-19 patients, providers should not rely solely on imaging to rule out infection. Instead, CXRs should be used in conjunction with clinical assessment, laboratory tests, and PCR testing when available.

What are the benefits of performing chest X-rays in urgent care settings for COVID-19 patients. Despite their limitations, CXRs can still provide valuable information:

Identifying alternative diagnoses (e.g., pneumonia, pulmonary edema)
Assessing disease severity in patients with abnormal findings
Establishing a baseline for monitoring disease progression
Guiding decisions about hospital admission or home management

Comparing Chest X-rays to Other Diagnostic Modalities

How do chest X-rays compare to CT scans and PCR testing in diagnosing COVID-19. Each modality has its strengths and limitations:

Chest X-rays

Widely available and relatively inexpensive
Lower sensitivity compared to CT scans
Normal findings do not rule out COVID-19

CT Scans

Higher sensitivity for detecting lung abnormalities
More expensive and less readily available
Concerns about equipment sterilization and radiation exposure

PCR Testing

Gold standard for diagnosing COVID-19
High specificity but variable sensitivity
Results may take hours to days, depending on laboratory capacity

Is there a role for combining multiple diagnostic modalities in COVID-19 evaluation. Integrating clinical assessment, laboratory tests, imaging studies, and PCR testing can provide a more comprehensive evaluation of patients with suspected COVID-19. This multi-modal approach may help overcome the limitations of individual diagnostic tools and improve overall diagnostic accuracy.

Optimizing Chest X-ray Utilization in COVID-19 Management

How can healthcare providers optimize the use of chest X-rays in managing COVID-19 patients. Consider the following strategies:

Use CXRs selectively, based on clinical presentation and risk factors
Interpret CXR findings in conjunction with other clinical and laboratory data
Recognize that normal CXRs do not exclude COVID-19 infection
Consider follow-up imaging for patients with worsening symptoms or high clinical suspicion despite normal initial CXRs
Implement appropriate infection control measures when performing and interpreting CXRs

What are the potential risks associated with relying too heavily on chest X-rays for COVID-19 diagnosis. Over-reliance on CXRs may lead to:

False reassurance in cases with normal findings
Delayed diagnosis and treatment for patients with COVID-19
Unnecessary radiation exposure if repeat imaging is performed frequently
Increased healthcare costs without proportional diagnostic benefit

Future Directions in COVID-19 Imaging Research

What areas of research should be prioritized to improve the utility of chest X-rays in COVID-19 management. Several avenues for future investigation include:

Developing AI-assisted interpretation algorithms to enhance CXR sensitivity
Investigating the prognostic value of CXR findings in predicting disease progression
Exploring the role of serial CXRs in monitoring treatment response
Evaluating the cost-effectiveness of different imaging strategies in various healthcare settings
Studying the long-term radiographic sequelae of COVID-19 infection

How might advancements in imaging technology impact COVID-19 diagnosis and management. Emerging technologies, such as portable X-ray devices and AI-enhanced image analysis, may improve the accessibility and accuracy of chest imaging in COVID-19 patients. These innovations could potentially bridge the gap between the convenience of CXRs and the sensitivity of CT scans, leading to more effective diagnostic strategies in urgent care and other ambulatory settings.

Documenting Chest X-ray Findings in Clinical Notes

How should healthcare providers document chest X-ray findings in patients with suspected or confirmed COVID-19. Accurate and comprehensive documentation is crucial for patient care, communication among healthcare team members, and medicolegal purposes. Consider including the following elements in your CXR documentation:

Key Components of CXR Documentation

Patient demographics and clinical context
Indication for the CXR
Technical quality of the image
Presence or absence of abnormalities
Description of any abnormal findings, including:
- Location (e.g., lower lobes, bilateral, multifocal)
- Pattern (e.g., interstitial changes, ground glass opacities)
- Extent (e.g., mild, moderate, severe)
Comparison to prior imaging studies, if available
Impression or conclusion
Recommendations for follow-up or additional imaging, if appropriate

What are some best practices for writing clear and concise CXR reports. Consider the following tips:

Use standardized terminology to describe findings
Organize the report in a logical, consistent format
Highlight the most significant findings
Avoid ambiguous language or hedging
Correlate imaging findings with clinical presentation
Provide a clear, actionable conclusion

How can healthcare providers improve communication of CXR findings to patients and families. Effective communication of imaging results is essential for patient understanding and shared decision-making. Consider these strategies:

Use plain language to explain findings
Provide visual aids or diagrams when appropriate
Emphasize the limitations of CXRs in COVID-19 diagnosis
Discuss the implications of findings on patient management
Address patient concerns and questions
Ensure proper documentation of the discussion in the medical record

Integrating Chest X-ray Findings into Clinical Decision-Making

How should healthcare providers incorporate CXR findings into their overall assessment and management of patients with suspected COVID-19. While CXRs provide valuable information, they should be considered in the context of the patient’s clinical presentation, laboratory results, and other diagnostic tests. Consider the following approach:

Steps for Integrating CXR Findings into Clinical Decision-Making

Assess pre-test probability based on clinical presentation and epidemiological factors
Interpret CXR findings in light of the patient’s symptoms and duration of illness
Consider alternative diagnoses that may explain the CXR findings
Evaluate the need for additional diagnostic tests (e.g., PCR, CT scan) based on CXR results and clinical suspicion
Use CXR findings to inform decisions about:
- Isolation precautions
- Treatment initiation
- Disposition (home management vs. hospital admission)
- Follow-up and monitoring plans
Communicate findings and management plans clearly to patients, families, and other healthcare team members

What are some potential pitfalls to avoid when interpreting CXR findings in the context of COVID-19. Be aware of the following challenges:

Over-reliance on normal CXRs to exclude COVID-19
Misattribution of abnormal findings to COVID-19 without considering other etiologies
Failure to recognize atypical presentations or co-infections
Neglecting to consider the timing of CXR in relation to symptom onset
Inadequate communication of findings and their implications to patients and colleagues

How can healthcare systems optimize the use of CXRs in managing the COVID-19 pandemic. Consider implementing the following strategies:

Develop clear guidelines for CXR utilization in COVID-19 evaluation
Provide education and training on CXR interpretation in the context of COVID-19
Implement quality assurance measures to monitor CXR utilization and interpretation
Establish protocols for infection control and equipment sterilization
Explore teleradiology options to facilitate rapid expert interpretation
Integrate CXR findings into electronic health records and clinical decision support tools

Chest X-Ray Findings in 636 Ambulatory Patients with COVID-19 Presenting to an Urgent Care Center: A Normal Chest X-Ray Is no Guarantee

Michael B. Weinstock, MD, Ana Echenique, MD, DABR, Joshua W. Russell, MD, MSc, FACEP, Ari Leib, MD, Jordan A. Miller, DO, David J. Cohen, MD, Stephen Waite, MD, Allen Frye, NP, and Frank A. Illuzzi, MD, FACEP

Abstract

Background/Objective

Patients with COVID-19 commonly present to Urgent Care (UC) centers. Our primary objective was to determine what percentage of UC patients with confirmed COVID-19 had normal vs abnormal chest x-rays (CXR). Secondarily, we aim to describe specific imaging characteristics and the frequency of each abnormal findings on plain film radiography (CXR).

Methods

A database of a large UC company in the greater New York City (NYC) area was reviewed for patients with positive SARS-CoV-2 PCR tests who also underwent CXR in UC between March 9 and March 24, 2020. Eleven board-certified radiologists, with knowledge that they were only reading imaging studies of COVID-19 patients, were each given a subset of the CXRs with oral and written instructions to re-read the films while disregarding the initial reading. Their readings were classified as normal, mild, moderate, or severe disease. They subsequently characterized specific findings. Lastly, overreads were compared with the initial CXR reading.

Results

Of the 636 CXRs reviewed among patients with confirmed COVID-19, 363 were male (57.1%) and 273 were female (42.9%). Patient ages ranged from 18 to 90 years of age, with most (493 patients, or 77.5%) being 30─70 years old. There were 371 CXRs re-read as normal (58.3%). Of the 265 abnormal cases (41.7%), 195 demonstrated mild disease, 65 demonstrated moderate disease, and five demonstrated severe disease. Interstitial changes and ground glass opacities (GGO) were the predominant descriptive findings in 151 (23. 7%) and 120 (18.9%) of the total, respectively. Location of the abnormalities were in the lower lobe in 215 (33.8%), bilateral in 133 (20.9%), and multifocal in 154 (24.2%). Effusions and lymphadenopathy were uncommon.

Discussion

This is the first study to specifically explore CXR findings of patients with confirmed COVID-19 evaluated in a UC setting. The vast majority of patients (566/636) had either normal or only mildly abnormal CXRs (89%), despite being symptomatic enough to warrant imaging as determined by the treating UC provider.

Conclusion

CXRs obtained from confirmed and symptomatic COVID-19 patients presenting to the UC were normal in 58.3% of cases, and normal or only mildly abnormal in 89% of patients. When abnormal, the most common findings were present in the lower lobes and the pattern was interstitial and/or multifocal. Pleural effusions and lymphadenopathy were uncommon.

INTRODUCTION

COVID-19, a novel disease caused by the SARS-CoV-2 virus, rapidly became a pandemic in early 2020, resulting in considerable worldwide morbidity and mortality. ^1,2 During this outbreak, acute care clinicians have been striving to accurately diagnose and define its clinical features in order to provide the best care for afflicted patients and limit the spread of the disease.

Plain film radiography of the chest (CXR) is a relatively inexpensive and widely available diagnostic modality in urgent care (UC) centers. However, to date, there is little evidence describing the utility of CXR in identifying patients with suspected COVID-19. Early observational studies discussing characteristic patterns of radiographic findings have focused predominantly on the use of computed tomography (CT) imaging. While CT has demonstrated good-to-excellent sensitivity (56%-91%) for COVID-19 lung pathology, depending on the interval between symptoms and imaging, cost and practical considerations (eg, sterilization after use) limit its utility, especially for use among ambulatory patients.^3-6

As most patients with COVID-19 seem to have a mild course of respiratory illness, evaluations are most likely to take place in nonemergency department and nonhospital settings, such as UC centers. ¹ In such settings, CXR is by far the most widely available imaging modality.⁷ However, to date all published studies of thoracic imaging findings in patients with COVID-19 have focused on hospitalized patients.⁸ Among such patients, Wong, et al found that the initial CXR had a sensitivity of only 69% for any abnormality.³ One would expect because UC patients typically have less severe disease, that CXRs among such patients would have even lower sensitivity compared with hospitalized patients.

In the Wong, et al study, the most common radiographic features in confirmed COVID-19 patients were peripheral rounded consolidations, ground glass opacities (GGO), and pulmonary nodules. Distribution of the lung changes were more common in the lower zones and bilateral.³ Even in asymptomatic patients, radiographic progression of disease, from focal unilateral changes to diffuse GGO and consolidations, was observed.⁸ Pleural effusions were rare and were associated with an increased risk of poor outcome. ³ Overall, the imaging changes reported peaked on days 10-12 of illness.^3,8

Our primary objective in this study was to determine what percentage of patients with confirmed COVID-19 had normal vs abnormal CXRs. Secondarily, we aimed to describe the frequency of each specific type of abnormal finding on plain film radiography (CXR).

METHODS

The electronic medical record (EMR) database of a large UC network with over 100 centers in greater New York City (NYC) and New Jersey (NJ) was queried, and 718 patients who had tested positive for SARS-CoV-2 by PCR between March 9 and March 24, 2020 (during the time that greater NYC was the epicenter for COVID-19) were identified. The CXRs for these patients were initially divided among 14 board-certified radiologists. However, due to willingness and ability to participate due to difficulty with remote access, only 12 were able to participate in the study. These individuals were assigned approximately 50 CXRs each, except for two of the radiologists who reviewed an additional 50 CXRs each to make up for the two radiologists who were not able to participate, giving these two radiologists a total of approximately 100.

Most participants re-read and correctly resulted 47 to 100 films. However, one radiologist only read 12 films; these readings were excluded from this report because the number of cases was far below the contributions of the other participants, providing a total analyzed sample of 636 CXRs (Figure 1).

Figure 1. Flowchart of All Confirmed COVID-19 Patients Seen in the UC Centers from March 9 to 24, 2020 Who Also Underwent CXR

Participating radiologists were given oral and written instructions to first categorize films as normal, mild, moderate, or severe disease; for those classified as abnormal, they were asked to describe the specific findings. Initial CXR readings were part of these patients’ medical records, but the radiologists were instructed to ignore the initial reading when they re-read the images. Participating radiologists were informed that the CXRs were from patients with confirmed COVID-19.

RESULTS

Eleven board-certified radiologists re-read CXRs of patients with PCR confirmed COVID-19 from multiple UC centers in the greater NYC area. Most participants re-read from 47 to 100 films, providing a total sample of 636 CXRs. Of these, 363 were male (57.1%) and 273 female (42.9%). Patient ages ranged from 18 to 90 years of age, with 493 patients (77.5%) being in the age range of 30─70 years old (see Table 1 and Figure 2).

Table 1. Demographics of UC Patients with COVID-19 Whose CXRs Were Re-Read by the 11 Radiologists (N=636)
Gender	n (%)
Male	363 (57.1%)
Female	273 (42.9%)

Of the 636 CXRs included in this report, 371 were re-read as normal (58.3%). Of the 265 abnormal cases (41.7%), 195 were classified as mild disease, 65 were classified as moderate disease, and five were classified as severe disease. Interstitial changes and GGO were the predominant descriptive findings in 151 (23.7%) and 120 (18.9%) of the total, respectively. Location of the abnormalities were in the lower lobe in 215 (33.8%), bilateral in 133 (20.9%), and multifocal in 154 (24.2%). Effusions and lymphadenopathy were uncommon (see Table 2).

Table 2. Characteristics of the Radiographic Findings Reported by the Panel of 11 Radiologists Who Re-Read CXRs of COVID-19 Patients Seen in Greater NYC UC Centers from March 9 to 24, 2020. (N=636)
Radiologic properties	Categories	n (%) of total
Severity	Normal	371 (58.3%)
	Mild	195 (30.7%)
	Moderate	65 (10.2%)
	Severe	5 (0.8%)
Type of infiltrate	Interstitial	151 (23. 7%)
	Ground Glass Opacities (GGO)	120 (18.9%)
	Consolidation	34 (5.3%)
Location	Lower	215 (33.8%)
	Upper	128 (20.1%)
	Diffuse	6 (0.9%)
Focality	Multifocal	154 (24.2%)
Focality	Focal	71 (11.2%)
Laterality	Bilateral	133 (20.9%)
Centrality	Peripheral	225 (35.4%)
Centrality	Central	45 (7.1%)
Other	Effusions	2 (0.3%)
Other	Lymphadenopathy	2 (0.3%)

Note: Numbers do not add to 100% as some patients had more than one finding.

The original readings from the medical records classified 468/636 (73.6%) CXRs as normal. When the CXRs were re-read for this study with the knowledge that the patients had COVID-19, 97 of these initial readings were changed to abnormal and two patients who had an initial finding of “possible pneumonia” were changed to “normal. ”

Classification as normal or abnormal varied across the 11 radiologists who did the re-reads for this study. On the lower end, one participant classified 14% of CXRs as normal, while at the upper end another participant classified 86% of CXRs as normal. Most participants classified between 51% and 80% as normal.

Specific examples of CXR images are presented in Figures 3–6.

Figure 3. Multifocal mixed central and peripheral linear infiltrates extending out to lung periphery with superimposed ill-define patchy opacities at the bilateral lung bases. Lung apices spared. Overall low volume, study concerning for hypoventilation. (X-ray courtesy of Experity Teleradiology (www.experityhealth.com/teleradiology.)

Figure 4. Subtle unilateral ground glass opacity at inferior margin of peripheral right upper lobe abutting the minor fissure. This patient has subtle unilateral involvement. (X-ray courtesy of Experity Teleradiology (www. experityhealth.com/teleradiology.)

Figure 5. Severe bilateral involvement with ill-defined patchy consolidation at periphery of right upper lobe, ground glass opacity at peripheral left upper lobe and central infiltrates extending to the lung bases from the pulmonary hila. Small rounded patchy infiltrate also noted at right lung base. (X-ray courtesy of Experity Teleradiology (www.experityhealth.com/teleradiology.)

Figure 6. Hazy ill-defined opacity at lower aspect of right upper lobe as well as rounded patchy infiltrates in right lung base and periphery of left lung base. (X-ray courtesy of Experity Teleradiology (www.experityhealth.com/teleradiology.)

LIMITATIONS

Studies of this type are inherently limited due to their retrospective and observational nature. Additionally, only a single CXR series was obtained for each patient. Because patients presented at various phases of illness, it is impossible to know whether patients with normal CXRs at time of presentation developed radiographic findings later during their illness.

We did not have access to data regarding patients’ underlying health histories nor baseline CXRs, therefore it is unclear to what extent abnormalities identified may have reflected chronic pulmonary conditions. However, most patients (454, or 71.4%) were <60 years of age and healthy enough to present in an ambulatory care setting and, therefore, would be expected, with infrequent exception, to have normal baseline CXRs.

Regarding CXR interpretation, although the radiologists were instructed not to let the initial CXR read, or knowledge of COVID-19 diagnosis influence their interpretation, they were not blinded to this information and we cannot rule out that it might have had an impact on their classifications. The shift to classify more CXRs as abnormal during the re-read suggest that this might be so.

We also did not have any assessment of inter-rater reliability between radiologists on the re-reads. The difference in percentage of normal classification across participants suggests that clear individual differences do exist among radiologists. However, as our purpose was to show what findings would be reported for COVID-19 patients in a clinical setting, the variability in CXR classification serves to highlight the challenges in real-time assessment in such patients.

The initial CXRs were obtained at the clinical discretion of the treating provider. It is likely that variations in medical decision-making and CXR utilization among providers influenced the availability of CXRs available for analysis among patients confirmed to have COVID-19. The direction of any associated bias is difficult to predict because many factors (eg, number of patients waiting to be seen, patient expectations etc.) influence providers’ decisions about imaging in UC patients with respiratory complaints.

Finally, the radiologists re-read the available CXRs looking for known varieties of abnormalities. It is possible that there are indications of disease on CXRs related to COVID-19 that are not yet defined (as this is a novel illness) and, therefore, the radiologists might not be expected to identify them.

DISCUSSION

This report is the largest observational study to date examining plain film radiographic findings among patients with COVID-19 in an ambulatory care setting. The majority of COVID-19 patients who present in this setting show no identifiable abnormalities on standard CXR assessment.

Though chest CT has been shown to be more sensitive than CXR, CT is generally not available in ambulatory care settings. Additionally, after scanning a patient with suspected COVID-19, extensive cleaning and decontamination of a CT scanner is required, making routine use of CT is impractical. The CXR, on the other hand, is widely available assessment tool in UC centers and allows relatively rapid cleaning and turn-over between patients.

Recently thoracic imaging consensus guidelines in COVID-19 have been published by the Fleischner Society.⁹ In patients with mild clinical features, imaging is indicated after a positive viral test if the patient has risk factors for disease progression. In a patient with moderate to severe clinical features, imaging is indicated after a positive viral test if the patient is at risk for worsening of pulmonary status. If testing for COVID-19 is unavailable, imaging can determine if an alternative diagnosis is present (eg, lobar pneumonia) or, if findings suspicious for COVID-19 are revealed, can guide further workup.¹⁰

When present, the patterns of abnormal findings were similar to those reported in other series of hospitalized patients with COVID-19^3,8 with peripheral, multifocal, and lower lobe involvement and interstitial or ground glass appearance being the most common. Additionally, pleural effusions and lymphadenopathy were relatively rare findings, which is also consistent with existing studies of chest radiography in COVID-19 patients. ⁸ Interestingly, alveolar disease was only bilateral in 133 (20.9%) of the total 636 CXRs, much less than reported in the CT literature where it is seen in 82% of cases.¹⁰ This may be due to the difficulty of perceiving early ground glass opacities on plain radiography and/or ambulatory patients presenting earlier in the course of illness.

In future reports we hope to examine what clinical signs, medical history, and demographic characteristic are associated with normal and abnormal CXR readings in patients with COVID-19.

CONCLUSION

CXRs obtained from confirmed and symptomatic patients with COVID-19 presenting to UC centers were normal in 58.3% of the patients, and normal or only mildly abnormal in 89% of patients. When abnormal, the most common findings involved the lower lobes and presented with an interstitial and/or multifocal pattern. Pleural effusions and lymphadenopathy were uncommon.

(This study was IRB approved and granted waiver of consent and full waiver of HIPAA authorization. No funding was obtained for this study.)

Acknowledgments:

The authors would like to acknowledge the following who donated their time and expertise to contribute to this paper:

Alyson Kendrick, QA Manager-TRS, Experity Teleradiology

B Ross MD, Radiologist

D Pellei MD, Radiologist

K Frame MD, Radiologist

C Chalfant MD, Radiologist

M Chalfant MD, Radiologist

M Donner MD, Radiologist

S Pawar MD, Radiologist

S Kathuria MD, Radiologist

M Hill MD, Radiologist

E Noeth MD, Radiologist

References

Guan WJ, Ni ZY, Hu Y, et al. Clinical characteristics of coronavirus disease 2019 in China. New Engl J Med. February 28, 2020. [Epub ahead of print]
Wu C, Chen X, Cai Y, et al. Risk factors associated with acute respiratory distress syndrome and death in patients with coronavirus disease 2019 pneumonia in Wuhan, China. JAMA Inter Med. March 13, 2020. [Epub ahead of print]
Wong HYF, Lam HYS, Fong AH-T et al. Frequency and distribution of chest radiographic findings in COVID-19 positive patients. Radiology. March 27, 2019. [Epub ahead of print]
Xu X, Chengcheng Y, Jing Q, et al. Imaging and clinical features of patients with 2019 novel coronavirus SARS-CoV-2. Eur J Nucl Med Mol Imaging. 2020;47:1275–1280.
Ng MY, Lee EYP, Yang J, Yang J, et al. Imaging profile of the COVID-19 infection: radiologic findings and literature review. Radiol: Cardiothor Imag. org/10.1148/ryct.2020200034.
Bernheim A, Mei X, Huang M, et al. ChestCT findings in coronavirus disease-19 (COVID-19): relationship to duration of infection. Radiology. February 20, 2020. [Epub ahead of print]
Trends in Urgent Care Radiology July 1, 2018 to June 30, 2019. Experity Urgent Care Quarterly. Issue 09; Winter 2020.
Shi H, Han X, Jiang N, et al. Radiological findings from 81 patients with COVID-19 pneumonia in Wuhan, China: a descriptive study. Lancet Infect Dis. 2020;20(4):425-434.
Rubin GD, Ryerson CJ, Haramati LB, et al. The role of chest imaging in patient management during the COVID-19 pandemic: a multinational consensus statement from the Fleischner Society. Available at: https://pubs.rsna.org/doi/10.1148/radiol.2020201365. Published online April 7, 2020.
Zhao W, Zhong Z, Xie X, et al. Relation between chest CT findings and clinical conditions of coronavirus disease (COVID-19) pneumonia: a multicenter study. Am J Roentgenol. March 3, 2020. [Epub ahead of print]

Michael B. Weinstock, MD is Professor of Emergency Medicine, Adjunct, Department of Emergency Medicine, Wexner Medical Center at The Ohio State University; and Senior Editor, Clinical, JUCM. Ana Echenique MD DABR is Clinical Director, Quality Management, Experity Teleradiology. Joshua W. Russell, MD, MSc, FACEP is Supervising Physician, Legacy-GoHealth Urgent Care; Emergency Medicine Staff, PeaceHealth Health System; Associate Editor, Urgent Care RAP; and Editor-in-Chief, JUCM. Ari Leib, MD is Adjunct Clinical Assistant Professor Emergency Medicine Ohio University Heritage College of Osteopathic Medicine. Jordan Miller, DO is Junior Chief Emergency Medicine Resident. David J. Cohen, MD is Medical Director, Experity Teleradiology. Stephen Waite, MD is Clinical Associate Professor of Radiology, SUNY Downstate Medical Center, Brooklyn, NY. Allen Frye, NP is a clinician and research coordinator. Frank A. Illuzzi, MD, FACEP is Executive Vice President, Quality & Education, CityMD. The authors have no relevant financial relationships with any commercial interests.

(When referencing this article, please cite as follows: Weinstock MB, Echenique A, Russell JW, et al. Chest x-ray findings in 636 ambulatory patients with COVID-19 presenting to an urgent care center: a normal chest x-ray is no guarantee. J Urgent Care Med. 2020;14(7):13-18.

You can download a pdf of this article here

Michael B.

Weinstock, MD

Associate Program Director, Adena Emergency Medicine Residency; Director of Research and CME, Adena Health System; Professor of Emergency Medicine, Adjunct, Department of Emergency Medicine, Wexner Medical Center at The Ohio State University; Senior Clinical Editor, The Journal of Urgent Care Medicine (JUCM)

Chest Radiograph

Author: William Krantz, MD, West Virginia University School of Medicine, Department of Radiology
Editor: Matthew Tews, DO, MS, Medical College of Wisconsin, Department of Emergency Medicine

Objectives

At the end of this chapter, the student will be able to:

Identify details to verify when looking at radiologic imaging
Assess adequacy of chest radiographs
Develop a systematic approach to reading chest radiographs
Identify important signs to know when reading chest radiographs
Understand the importance of knowing common diagnoses on the chest radiograph

Introduction

Section 1: Verify Details

Although mistakes are rare, they do occur and you should verify details pertaining to your patient on every study you review.

Patient – Ensure the study was performed on the correct patient.
Time and date – Note the time and date of the study. This is especially important if there are comparison studies. By convention, the older study is displayed on the left, though this is not always the case.
Study type – Verify the type of study performed. Sometimes a study will be changed depending on what the patient can tolerate. For example an AP supine CXR may be done on a patient that can’t stand for a PA/lat CXR.
Comparisons – Make sure you view every image submitted. It is not uncommon for multiple views to be taken for an AP supine CXR study on a difficult to position patient. Compare these to previous studies, if available.

Projection

Section 2: Assess Adequacy

PA/lateral projection is the standard used for most patients who are ambulatory and able to stand. In this view, the mediastinum should have a normal width and a good inspiratory effort should results in full diaphragm expansion.

PA Projection – Normal Mediastinum and Good Inspiratory Effort

AP projection is obtained with the patient in bed and lying flat or partly upright. It is usually reserved for non-ambulatory patients. AP projections have the disadvantage of making the heart an mediastinum appear more prominent as well as usually resulting in shallower inspiration which can limit evaluation of the lung bases. A lateral projection is not obtained with a AP projection, thereby limiting the view of the chest further.

AP Projection – Widened Mediastinum and Less Inspiratory Effort

It is therefore preferable to order a PA/lateral CXR over an AP view when possible.

Rotation

It is not uncommon, especially for AP supine CXRs, for the patient to be slightly rotated. Rotation can be assessed by measuring the distance between the medial edges of the clavicles to the vertebral spinous processes. They should be equal or near equal. Anterior structures move the same direction as rotation so the clavicle/spinous process width is increased on the side to which the patient is rotated. This patient is rotated to the left.

Red Dotted Line = Slight Rotation to the Left

Here is a patient severely rotated to the left. Severe rotation can alter the normal cardiomedistinal contour and make interpretation difficult.

Significant Rotation to the Left

Inspiration

Deeper inspirations show more lung and result in better overall images with less haziness at the lung bases and less enlargement of the heart and mediastinum. A good inspiration on a PA CXR shows at least 9 posterior ribs.

The following films were of the same patient and taken using the same AP projection. The image on the left is a poor inspiratory effort (ribs 1-6), while in the film on the right, the patient achieved a much deeper inspiration on the bottom xray (ribs 1-10).

Poor Inspiratory Effort Good Inspiratory Effort

Penetration

Under-penetration results from not enough xrays passing thru to allow differentiation of dense structures, thus the mediastinum and spine appear white. In an over-penetrated CXR too many xrays have passed thru the chest preventing differentiation of low density structures, thus the lung fields appear black. With proper penetration, the spine should be faintly visible behind the heart.

Under-penetrated Over-penetrated Correctly-penetrated

Section 3: Develop Systematic Approach

It is important to develop a system for looking at imaging and be consistent in its use. Different systems exist, whether it is based on anatomical structure (looking at one structure, then another, and so on) or scanning the image in a pattern (back and forth from top to bottom). No system is best. What is important is finding one that you like and using it consistently. Resist the natural tendency to focus in on the area which first catches your eye otherwise you may miss the more important subtle finding.

ABCDs

The ABCDs approach is taught as an aid for remembering steps in many different learning environments. It can be applied in interpreting chest xrays too. The mneumonic starts with A and can go through H.

A – Airway	E – Effusion
B – Bones	F – Fields
C – Cardiac	G – Gastric
D – Diaphragm	H – Hila

A – Airway

Identify trachea and note if it is midline and straight. A deviated trachea can be caused by many etiologies including rotation and masses.
Identify carina and estimate the angle. Normally 60-100°. Common things that may increase this angle include left atrial enlargement and adenopathy.
Trace out right and left mainstem bronchi.

Trachea midline: yellow; Spinous process and medial clavicles: red. This patient is slightly rotated right.

Leftward tracheal deviation due to large substernal goiter

Endotracheal tubes

Endotracheal tube (ETT) tip should sit 3-5cm above the carina. The ETT can be too high above the carina and will need to be advanced, or too deep into the right mainstem and need to be pulled back.

Intubation with Good Position of ETT

Red = ETT; Yellow = Trachea and Bronchi

Right mainstem bronchus intubation

B – Bones

Evaluate clavicles, AC joints, GH joints and humeri followed by the ribs and vertebra.

Normal Alignment of Clavicle, Glenohumeral Joint and Proximal Humerus on a PA View of Chest

Subtle Anterior Shoulder Dislocation.

The humeral head (solid yellow line) is out of the glenoid fossa (dotted yellow line)

Sometimes, overlying structures on the radiograph can obscure positive findings. If there is suspicion of an injury or abnormality in the region of overlying structures (such as cords, belts, jewelry), remove them and repeat the radiograph.

“Normal” Chest Radiograph at First Glance

However, cardiac monitor cords overly the distal clavicle

Distal Clavicle Fracture

The film shows a distal clavicle fracture when the monitor cord is removed

Rib fractures can sometimes be extremely subtle. Be sure to trace all the ribs to identify step offs.

Left Sided Upper Rib Fractures

A useful trick can be to flip the image upside down or sideways which makes the ribs appear to stand out.

Left Sided Rib Fractures

Visualize the entire spine and look for vertebral body height loss and alignment

L1 Compression Fracture = Red; T11 and T12 Normal Body Height (Yellow)

C – Cardiac

Next, check the cardiac size: normal is <50% of thoracic diameter on PA projection and <55% on AP projection.

Normal Cardiac Size on PA CXR

Red = thoracic diameter; Yellow = cardiac diameter

Verify cardiac borders are sharp and defined.

Evaluate aorta and AP window (should be concave)

Normal AP Window and Cardiac Border

Red = Aortic contour; Blue = Upper cardiac border; Yellow = AP window

D – Diaphragm

Verify diaphragm and costophrenic angles are sharp and well defined.

Sharp Diaphragmatic Angles = Yellow

Note any flattening of diaphragm which may indicate hyperexpanded lungs such as seen in a patient with COPD.

Flattened Diaphragm in COPD = Yellow

Obscuration of all or a portion may indicate pleural effusion or consolidation.

Left Sided Pleural Effusion

Look for free air under diaphragm on upright projections.

Free Air Under Diaphragm = Yellow

E – Effusion

Look for ‘blunting’ of the costophrenic angles on PA projection and posterior costophrenic sulcus on lateral projection. As much as 300-500cc of fluid may not be seen on PA projection, but seen on lateral projection.

Pleural Effusion on Lateral View

Fluid may track into right minor fissure causing it to thicken or even form a pseudotumor.

Right Minor Fissure Fluid

Loculated effusion may present as nondependent pleural thickening.

Right Sided Loculated Effusion

F – Fields

Lung fields should generally be uniformly dark grey.

Uniformly Dark Grey Lung Fields

Increased density may be due to airspace disease such as pneumonia, masses, pleural effusions, pulmonary edema, or atelectasis

Left Lower Lung with Effusion

Decreased density may be due to pneumothorax or hyperexpansion due to COPD.

Left Sided Pneumothorax with Tension Component

Heart is being pushed to the right

There are several other findings to look for with the lung fields. Sometimes they are obvious and take over most of the view of the lung fields and other times they are subtle and scattered or singular.

‘Bat-wing’ Pulmonary Edema

Septic Emboli

G – Gastric

Look for gastric air bubble located in LUQ below the left hemidiaphragm and cardiac silhouette.

Hiatal hernias are very common. Usually seen as a rounded density at the midline behind the heart.

Sometimes an air-fluid level may be seen.

Diaphragmatic Hernia

Kartagener’s Disease (dextrocardia, airspace disease, situs inversus)

A case that at first glance appears to represent massive cardiomegaly, but closer inspection of the lateral view reveals air density anteriorly that resembles air in the bowel. Finding was suspicious for hernia.

Coronal CT image reveals mesenteric fat and bowel in a large diaphragmatic hernia.

H – Hila

Locate right and left main pulmonary arteries.

Note any obvious contour abnormalities which may indicate masses or adenopathy.

Normal Pulmonary Artery Contour

Section 4: Learn Important Signs

Silhouette Sign

Normally on Xray studies, two adjacent structures are visually distinct from one another because they have different densities. Thus you can see a silhouette of the dense structure against the adjacent less dense structure, for example the cardiac silhouette or diaphragms adjacent to the air containing lungs.

Normal Cardiac Silhouette

If something alters the tissue density, typically the less dense lung becoming more dense due to mass or consolidation, the silhouette may be lost because the adjacent structures are now closer in density.

Left Sided Effusion Causing Loss of Cardiac Silhouette

The classic example is a right middle lobe (RML) consolidation. Normally, the right heart border silhouette is seen because it is next to air containing lung. In the presence of a RML consolidation which replaces air with denser inflammatory material, the right heart silhouette is lost.

Normal Right Heart Border Loss of Right Heart Border from RML Infiltrate

A right lower lobe consolidation will not result in loss of right heart silhouette because they are not adjacent to each other.

RLL Consolidation – No Loss of Right Heart Border

Sometimes it is not obvious on the PA view which side a consolidation is on because it may be located in the posterior sulcus and therefore obscured.

We can use the lateral view and an understanding of the silhouette sign to help determine when a consolidation is present and which side it is on.

“Normal” View of PA and Lateral Chest

Notice how on this normal lateral the diaphragms extend all the way from anterior to posterior. The right diaphragm usually is slightly higher than the left and the left often will not extend fully anterior because of the heart being adjacent to it.

Now notice how the left diaphragm silhouette no longer extends posteriorly all the way. This is because it has lost its silhouette because there is a dense consolidation directly adjacent to it.

Loss of Left Diaphragm Due to Consolidation

Air Bronchogram

Normally bronchioles are not visible because they are surrounded by air in adjacent alveoli

In the setting of air in the alveoli being displaced by fluid or cellular debris, the air in the bronchioles appears as a translucent tube against the hazy opacity of the affected airspaces.

This can be a clue that a opacity may be due to airspace consolidation and not mass.

Air bronchogram sign (enlarged right lower lobe image on right)

Deep Sulcus Sign

In patients who are upright when imaged, such as for a PA/lateral CXR, air in the pleural space from a pneumothorax tends to collect in non-dependent locations, such as the apices.

Left Apical Pneumothorax

In patients who are supine, air may collect at the bases and anterior chest, potentially resulting in a deep sulcus sign, which indicates a pneumothorax.

Deep Sulcus on Right (Yellow Arrows) on Supine Radiograph

Continuous Diaphragm Sign

Normally, the central portion of the diaphragm is not visible because it is contiguous with the cardiac silhouette.

“Normal” Appearance of the Right and Left Diaphragms

If it is visible then it is highly suggestive of free air in the mediastinum or peritoneal cavity.

Continuous Diaphragm from Intraperitoneal Free Air

(Dotted yellow line outlines mediastinal air)

References

Brant, William E, and Clyde Helms. Fundamentals of Diagnostic Radiology, 4th ed. Philadelphia: Lippincott Williams and Wilkins, 2012. Hardback.
Corne, Jonathan, and Maruti Kumaran. Chest X-Ray Made Easy, 4th ed. Edinburgh: Elsevier, 2015. Paperback.
Guttentag, Adam. “Basic X-ray Interpretation,” Learning Radiology,http://www.learningradiology.com/lectures/facultylectures/Basic%20Chest%20X-Ray%20Interpretation/player.html.
Radiology Masterclass, http://www.radiologymasterclass.co.uk/tutorials/tutorials
Smithuis, Ron and Otto van Delden. “Chest X-Ray – Basic Interpretation,” The Radiology Assistant, http://www.radiologyassistant.nl/en/p497b2a265d96d/chest-x-ray-basic-interpretation.html.

An example of OpenI [2] chest x-ray image, report, and annotations.

Context 1

… publicly available radiology dataset is ex- ploited which contains chest x-ray images and reports pub- lished on the Web as a part of the OpenI [2] open source literature and biomedical image collections. An example of a chest x-ray image, report, and annotations available on OpenI is shown in Figure 1. …

Context 2

… ever, a few findings have been rendered uninterpretable. More details about the dataset and the anonymization pro- cedure can be found in [11], and an example case of the dataset is shown in Figure 1. …

Context 3

… report is structured as comparison, indication, find- ings, and impression sections, in line with a common radi- ology reporting format for diagnostic chest x-rays. In the example shown in Figure 1, we observe an error resulting from the aggressive automated de-identification scheme. A word possibly indicating a disease was falsely detected as a personal information, and was thereby “anonymized” as “XXXX”. …

Context 4

… radiology reports contain comprehensive information about the image and the patient, they may also contain information that cannot be inferred from the image content. For instance, in the example shown in Figure 1, it is probably impossible to determine that the image is of a Burmese male. On the other hand, a manual annotation of MEDLINE R citations with controlled vocabulary terms (Medical Subject Headings (MeSH R ) [1]) is known to significantly improve the quality of the image retrieval results [20,22,10]. …

Context 5

… annotation generation examples are provided in Figures 10 and 11. Overall, the system generates promis- ing results on predicting disease (labels) and its context (attributes) in the images. …

Context 6

… rare disease cases are more difficult to detect. For example, the cases pul- monary atelectasis, spondylosis, and density (Figure 10), as well as foreign bodies, atherosclerosis, costophrenic angle, deformity ( Figure 11) are much rarer in the data than cal- cified granuloma, cardiomegaly, and all the frequent cases listed in Table 1 of the main paper. …

Context 7

Context 8

… the (left or right) location of the disease cannot be identified in a lateral view (obtained by scanning the patient from the side), as shown in Figure 11. Since our dataset contains a limited number of disease cases, we treat each x-ray image and report as a sample, and do not account for different views. …

Automated abnormality classification of chest radiographs using deep convolutional neural networks

Our study was compliant with the Health Insurance Portability and Accountability Act and was conducted with approval from the National Institutes of Health Institutional Review Board (IRB) for the National Institutes of Health (NIH) data (Protocol Number: 03-CC-0128, Clinical Trials Number: NCT00057252), and exemption from IRB review for Indiana and Guangzhou datasets. The requirement for informed consent was waived.

Datasets

We studied three different databases. 1. National Institutes of Health Database: two subsets were used from this database: (a) NIH “ChestX-ray 14” dataset: A total of 112,120 frontal-view chest radiographs and their corresponding text reports were obtained retrospectively from the clinical PACS database at the NIH Clinical Center. We text-mined the radiological reports using the same Natural Language Processing (NLP) techniques used in the ref. ¹⁶. The abnormalities of major abnormal cardiac and pulmonary findings in this dataset include cardiomegaly, lung opacity (including pneumonia, consolidation, and infiltrate), mass, nodule, pneumothorax, pulmonary atelectasis, edema, emphysema, fibrosis, hernia, pleural effusion, and thickening. These abnormalities were binned into the “abnormal” category, and negative studies were included in the “normal” category. Note that the patients with medical devices (e.g., chest tubes, central venous catheters, endotracheal tubes, feeding tubes, and pacemakers) or healed rib fractures but without any other chest abnormalities were categorized into the “normal” category. We approximately balanced the “normal” and “abnormal” categories (about 50% for each category) to ease the training and evaluation procedures. After automated NLP mining, a total number of 11,596 radiographs were obtained, among which 10,252 were separated into training and validation sets and 1344 for hold-out testing. The labels for the training and validation sets were obtained using only the automated NLP tool, while two different sets of labels were obtained for the testing set. The first set of labels were obtained by using the same NLP tool as above and then corrected by an expert based on the radiology reports. More specifically, a “human in the loop” manual correction process was applied on the 1344 testing images and reports. This process was adopted to correct some potential wrong labels extracted using NLP, from the text reports composed by the attending radiologists. In this process, a human observer (Y.X.T.) checked the label consistency between the binary NLP label and the impression (conclusion) of the attending radiologist, which indicates if a chest X-ray is normal or abnormal in the text report. If there was a discrepancy, a radiologist (M.B.) read the text report and drew conclusion (normal or abnormal). 33 images were sent to the radiologist and 26 of them were eventually corrected by the radiologist. This indicates that the accuracy of NLP on the binary labeling task is 98.07%. This is the so-called “attending radiologist label set”. The other set of labels was obtained by taking the consensus of three US board-certified radiologists. This is denoted as “consensus of radiologists label set”. 677 images were labeled as normal and 667 images were labeled as abnormal by the attending radiologist, while 691 images were labeled as normal and 653 were labeled as abnormal by the consensus of three radiologists. We perform seven-fold cross-validation (about 8500 images for training and the rest for validation) and report the mean and standard deviation results in this experiment. (b) RSNA pneumonia detection challenge dataset: a total of 25,684 chest radiographs from the NIH database were re-labeled by six board-certified radiologists from the Radiological Society of North America (RSNA) and two radiologists of the Society of Thoracic Radiology (STR) into three categories: normal (n = 8525, 33.2%), abnormal with lung opacity (n = 5659, 22.0%) and abnormal without lung opacity (n = 11,500, 44.8%). The definition of “pneumonia-like lung opacity” includes findings like pneumonia, infiltration, consolidation, and other lung opacities that radiologists considered as pneumonia-related. The details of the dataset and annotation process can be found in the ref. ³³. 2. Indiana University Hospital network database: we used the chest radiographs from the Indiana University hospital network publicly available at the Open-i service of the National Library of Medicine. This dataset contains chest radiographs obtained in both the frontal and lateral projections. We trained an automated tool (available at https://github.com/rsummers11/CADLab) to classify the two views and filtered 3813 de-identified frontal chest radiographs, among which 432 (50% normal, 50% abnormal) were used for testing. The remaining radiographs were used to fine-tune the model trained on the NIH “ChestX-ray 14” dataset. 3. Guangzhou Women and Children’s Medical Center Pediatric Database: a database from Guangzhou Women and Children’s Medical Center (WCMC) in China containing 5856 pediatric chest radiographs were made publicly available by Kermany et al. ³⁴. Chest radiographs in this database were either labeled as normal or pneumonia (caused by virus or bacteria). We used the same data split as in the ref. ³⁴, where 5232 (1349 normal, 3883 pneumonia) images were used for training and validation, and the remaining 624 (234 normal, 390 pneumonia) radiographs were used for testing.

Deep convolutional neural network structure and development

We trained various well-known deep CNN architectures such as AlexNet²⁵, VGGNet²⁶, Inception-v3 (GoogLeNet)²⁷, ResNet²⁸, and DenseNet²⁹. The weights (or parameters) of these models were either pre-trained on about 1.3 million natural images of 1000 object classes from the ImageNet Large Scale Visual Recognition Challenge database³⁰ (the so-called “transfer learning” strategy) or randomly initialized (the so-called “training from scratch” strategy). We replaced the final classification layer (1000-way softmax) of each pre-trained CNN with a single neuron with sigmoid operation that outputs the approximate probability that an input image is abnormal. We resized each input chest radiograph to 256 × 256, cropped 224 × 224 center pixels (for Inception-v3, we resized the image to 342 × 342 and cropped 299 × 299 center pixels in order to make it compatible with its original dimensions), and fed them to each individual CNN model. We also evaluate with different input radiograph sizes such as 512 × 512 (448 × 448 crop) and 1024 × 1024 (896 × 896 crop) pixels. CNN models were trained using backpropagation on an NVIDIA TITAN X Pascal graphics processing unit (GPU) with 12 GB memory for 256 × 256 images and on an NVIDIA TITAN V-100 GPU with 32 GB memory for 512 × 512 and 1024 × 1024 images. The loss function was binary cross-entropy loss. We used a grid search to find optimal hyperparameters (learning rate, batch size, etc.). All the layers of the ImageNet pre-trained CNN models were fine-tuned using an initial learning rate [0.0005, 0.001, 0.05, and 0.1] ([0.005, 0.01, 0.05, and 0.1] for models with random initialization) with a weight decay rate of 0.0001, using the stochastic gradient descent (SGD) optimizer with the momentum of 0.9. The learning rate was reduced by a factor of 0.1 after the loss plateaued for five epochs. Early stopping was used to avoid overfitting on the training set with a maximum running of 50 epochs. The batch size was [64, 128] for an image size of 256 × 256, [16, 32] for 512 × 512 and [4, 8] for 1024 × 1024. We empirically found for 256 × 256 input images and a batch size of 64, the optimal learning rate was 0.001 for ImageNet pre-trained models and 0.01 for models with random initialization. We augmented the dataset in the training stage by horizontally flipping the chest radiographs. We implemented the networks using the open-source PyTorch (https://pytorch.org/) deep learning framework.

Reader study

Four radiologists (Radiologist #1, #2, and #3 are US board-certified, Radiologist #4 is a foreign-trained radiologist) served as human readers to label the same NIH “ChestX-ray 14” test set above. They had a mean of 29.75 years of experience (range 29–31 years). Annotation was performed by using a customized graphical user interface (GUI)-based annotation software installed on readers’ personal computers. The readers were shown chest X-rays in Portable Network Graphics (PNG) format with an image size of 1024 × 1024 pixels; they were able to zoom in and out using the software. The readers were provided with the same guidelines to the annotation software and rules. They were to make binary decisions on the 1344 chest radiographs and were blinded to the text report composed by the attending radiologist who read the original scan and other readers’ annotations. The ratio of normal to abnormal radiographs was not revealed to the readers.

Quantification and statistical analysis

The predictive performance of the deep CNN models was compared with that of practicing radiologists. We performed seven-fold cross-validation on the training and validation subsets and averaged outputs (scores) of seven models on the test set. The performance metrics were the AUC, sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), F1 score, accuracy, and confusion matrix. The 95% confidence intervals (CI) were obtained using seven-fold cross-validation. Cohen’s kappa coefficient³⁵ was used to assess the inter-rater agreement. These measurements were computed using scikit-learn (https://scikit-learn.org), a free software machine learning library for the Python programing language (https://www.python.org/). The ROC curves were plotted using matplotlib (https://matplotlib.org/), a plotting library for Python. Note that the computer program gave an approximate probability that a chest radiograph was abnormal, while the radiologist only provided a binary (normal or abnormal) decision on a chest radiograph. We set a hard threshold to 0.5 to determine the binary decision of the computer program when required in computing the metrics. Comparisons between AUCs were obtained by using a nonparametric approach³⁶, where multiple replicates of each model were trained and tested. We used a t-test, provided by the ttest_ind function in SciPy (https://www.scipy.org/), an open-source Python library for scientific computing and technical computing, for the statistical test, with a P-value less than 0.05 indicating statistical significance. Qualitative results were visualized by highlighting the image regions that were most responsible for the deep CNN classification model using class activation maps^4,16,31.

Artificial Intelligence Shows Potential for Triaging Chest X-rays

An artificial intelligence (AI) system can interpret and prioritize abnormal chest X-rays with critical findings, potentially reducing the backlog of exams and bringing urgently needed care to patients more quickly, according to a study appearing in Radiology.

“Currently there are no systematic and automated ways to triage chest X-rays and bring those with critical and urgent findings to the top of the reporting pile,” said study co-author Giovanni Montana, PhD, formerly of King’s College London and currently at the University of Warwick in Coventry, England.

In the U.K. there are an estimated 330,000 X-rays at any given time that have been waiting more than 30 days for a report. Deep learning (DL) has been proposed as an automated means to reduce this backlog and identify exams that merit immediate attention, particularly in publicly-funded health care systems.

AI System Developed to Identify Key Findings

For the study, Professor Montana and colleagues used 470,388 adult chest X-rays to develop an AI system that could identify key findings. The images had been stripped of any identifying information to protect patient privacy. The radiologic reports were pre-processed using Natural Language Processing (NLP). For each X-ray, the researchers’ in-house system required a list of labels indicating which specific abnormalities were visible on the image.

“The NLP goes well beyond pattern matching,” Dr. Montana said. “It uses AI techniques to infer the structure of each written sentence; for instance, it identifies the presence of clinical findings and body locations and their relationships. The development of the NLP system for labeling chest X-rays at scale was a critical milestone in our study.”

The NLP analyzed the radiologic report to prioritize each image as critical, urgent, non-urgent or normal. An AI system for computer vision was then trained using labeled X-ray images to predict the clinical priority from appearances only. The researchers tested the system’s performance for prioritization in a simulation using an independent set of 15,887 images.

Immediate reporting of chest X-rays referred from general practice by reporting radiographers: a single centre feasibility study

https://doi.org/10.1016/j.crad.2017.11.016Get rights and content

Highlights

•: Early lung cancer diagnosis is often limited by insufficient radiology capacity.
•: It is feasible to introduce immediate reporting of chest X-rays from general practice by radiographers.
•: Time to diagnosis of lung cancer can be significantly shortened with immediate chest X-ray reporting.

Aim

To investigate the feasibility of radiographer-led immediate reporting of chest radiographs (CXRs) referred from general practice.

Materials and methods

This 4-month feasibility study (November 2016 to March 2017) was carried out in a single radiology department at an acute general hospital. Comparison was made between CXRs that received an immediate and routine report to determine the number of lung cancers diagnosed, time to diagnosis of lung cancer, time to computed tomography (CT), and number of urgent referrals to respiratory medicine.

Results

Forty of 186 sessions (22%) were covered by radiographer immediate reporting. Of the 1,687 CXRs referred from general practice, 558 (33.1%) received an immediate report (radiographer or radiologist). Twenty-two (of 36) CT examinations performed were following an abnormal CXR with an immediate report (mean 0.8 scans/week). Time from CXR to CT was shorter in the immediate report group (n=22 mean 0.9 days SD=2.3) compared to routine reporting (n=14; mean 6.5 SD=3.2; F=27.883, p<0.0001). Time to multidisciplinary team (MDT) discussion was shorter in the immediate reporting group (mean 4.1 SD=2.9) compared to routine reporting (mean 10.6; SD=4.5; F=11.59, p<0.0001). No apparent difference was found for time to discussion at treatment MDT.

Conclusion

It is feasible to introduce a radiographer-led immediate CXR reporting service. Patients can be taken off the lung cancer pathway sooner with the introduction of radiographer immediate reporting of CXRs and this may improve outcomes for patients. A definitive study assessing outcomes is required to determine whether this will have an impact mortality and morbidity for patients.

Recommended articlesCiting articles (0)

Citing articles

Chest X-Ray Findings Normal for Many Confirmed COVID-19 Cases

Chest X-Ray Findings Normal for Many Confirmed COVID-19 Cases | PracticeUpdate

center of excellencealarmbookmarkcalendarcheck-clipboardclockwisecompass-navigatecompasscounterclockwisedeletedownload-bookeducation-historyemailget-helpgizmo-flaghelp-iconjournalleftlinklocklog-inlog-outnavigate-downnavigate-leftnavigate-rightnavigate-upnotifications-disabledopen-bookpersonplay-videoplayrefreshrightrightwardsearchsettingsshare-2starthumbs-upsharecheck-box3linesicon-social-facebookicon-social-instagramicon-social-linkedinicon-social-twittericon-social-youtubeprinter-2insert-boxeshelp-iconlocationpluspaperclipdownloadclockwisecheck-circleexclamation

We have detected that you are using an Ad Blocker.

PracticeUpdate is free to end users but we rely on advertising to fund our site. Please consider supporting PracticeUpdate by whitelisting us in your ad blocker.

We have sent a message to the email address you have provided, . If this email is not correct, please update your settings with your correct address.

The email address you provided during registration, , does not appear to be valid. Please update your settings with a valid address before to continue using PracticeUpdate.

Sign in to PracticeUpdate

Only registered members have full access to PracticeUpdate content.

Denis and I are one combat unit of the Distance team. To complete the combat mission, we have everything you need: a marking tape, scissors, cans of paint for asphalt, a sea of signs and signs, poles, colored tape, construction staplers and you never know what else. Lanterns, some food, isotonic. I also always have matches with me – you never know when they will come in handy, but if at that moment they are not there, the problems will be more serious than a callus on the tip of my little finger.We know how to make the eyes of the participants, from which the river flows (no, not tears) lactate, cling to our “bread crumbs” – marking tape. Our needs have been reduced to a minimum. We want to eat and sleep all the time. Somewhere out there in civilization, at the start or finish, Sasha is waiting for me, and I have an interesting Final of the Race in stock. This is how we work.

Not even the Doctor complained this time. I’m used to it, apparently. Few of whom I have heard during the race that this is all impossible.Everyone has already understood that this is how steep races are held in Russia, and we will not be able to work otherwise. It turns out that you can even get used to time trouble. The payback comes later when we relax. The trouble is that there will be no time to relax.

But, as usual, let’s try it in order.

Chapter 1. Before the race.

The road to Crimea generally started out great: Sasha and I drove from Moscow by car, enjoyed the road trip, tasting the dubious delights of spending the night in a roadside motel, eating at the M sports food restaurant and at gas stations, casual communication with traffic police officers on the topic of movement on the lane of oncoming traffic and unexpected delays in the operation of the crossing.The reward for the difficulties was a wonderful road from Feodosia to Sudak along the sea and a warm and abundantly delicious welcome from Sasha’s grandmother.

The next couple of days we were just resting. We ran through the surrounding mountains, and I also carried out an important part of the preparation for the planned Final of the Race. After resting and restoring peace of mind, it’s time to get ready for the journey. We leave on Monday afternoon towards Balaklava. From there, everything will begin on Wednesday, 580 hot heels (read – 290 singing hearts) will start at one of the most beautiful and at the same time difficult tests of the season – the Crimea X Run race.

I must admit, for the race most of all were done by two people, without whom nothing would have happened: 1. Ivan Petrov. This person is not just the heart and limbs of CXR, but also our Leader. Vanya is 60% of CXR’s success and nothing less. He’s our Control Packet, twix and a half sticks. 2. Ilyukha Tsygankov. Just Ilyukha. He is the megamind of our campaign. Designs, texts, calculating control times, creating booklets for participants and volunteers in each area of responsibility – his business. I can’t imagine how many hours lie in the foundation of this work.Ilyukha is 60% of the coolness and beauty of our project. He is the “hidden Steve Jobs” of the CXR. If there were others in their place, the CXR would be completely different. They are the soul and character of the CXR.

This does not mean that the rest of the “main backbone” somehow did little. Dasha. As many questions as she handles, only one person could manage, besides her – Ivan Petrov. But cloning people is still prohibited, so it’s very cool that there is Dasha in the world. Next is Oleg. I don’t understand how you can collect all the racing “shmurdyak” (the working name of the huge amount of equipment that is used in the race) in Moscow, organize logistics and endless assemblies of start and finish towns and preparation of sites for evening events.This is also one of the terribly capacious tasks. And further down the list – the secretariat, Masha, Natasha and their CO. Compared to all of them, our distance services are just easy and enjoyable cross-country hikes.

I have no words to describe how I respect these people. As Sasha Chervyakov said, organizing races is like cycling, the main thing is to learn, only the bike is on fire, and you are on fire, and everything around is on fire. It’s true, and I’m always happy to burn with these guys. Well, okay, back to the race.

Chapter 2. Day zero.

We don’t have to run anywhere on Monday. It’s even strange. No need to scout a new track for the last day and wander around looking for options. An unusual sensation. Well, nothing, so we’ll snatch it somewhere further. The commotion began the next morning, Tuesday. We were unloading a lorry, going along the way for the marking of the first day. It was decided WHAT and WHEN we mark, but of a very general nature. To begin with, Denis and I ran with ribbons for the second part of the first day’s route, leaving the car at the food point, while Tyoma and Blagov (aka Sasha, but everyone knows him as Blagova) ran the first part and picked up the car.A well-developed, proven scheme.

We work, as usual, smoothly, but in a training mode. For Denis, this is the first experience of marking and he has yet to understand why it is necessary to hang the markings on the branch of a bush or tree closest to the participants, visible from afar, how to mark the turns, and just what needs to be done so that the participants are less worried and run away to less necessary. We cannot completely exclude this, otherwise we would have had to pull the corridor out of the tape, which is 320 kilometers of marking.But even then, I am sure, there would have been at least one person who ran out of this corridor.

But we ran well, it came out about 15.5 kilometers with 750 meters of recruitment. We climbed the rocks of the Temple of the Sun, admired the plateau from the top of Chelyabi and fled along a narrow path to Foros. For a warm-up it is the most. True, the backpacks weighed somehow not humanely, but we will get used to it.

In the evening we have an opening. How many people go crazy! Sasha has already been turned into a “hook”, and she doesn’t mind at all.We are working! They talked about the distance, tried to livelier and without snot.

Chapter 3. Day One.

Our days start at least 2 hours before the start of the participants. Today with Denis we are running the first part of today’s route – what the guys marked yesterday. Already yesterday, a battle for the markings unfolded on this site: Sveta, a participant in all our “fornicators” from races to camps, went for a walk and neutralized the “crazy grandmother” who was tearing down our markings. But in the morning it turned out that the granny, most likely, had accomplices, and the markings, which Sveta had restored yesterday, were again torn off.We have dramatically increased work. This means that today we are doing short intervals: acceleration for 10-20 seconds, then a knot, turning the tape with a picture towards the participants and again accelerating. And so many, many times. It turned out that they removed about 7 kilometers, and then it went easier. Denis understands that we cannot lose, this is not a race where you can give up the slack. But we made it in time, I even took a couple of photos at the top of Kush-Kai. I run and envy the participants who do not set ambitious goals for themselves. This is who will definitely enjoy the views, forests, access to the plateau and watermelons at the food station.

By the way, here he is. I love to scare the PC workers because they think the members are running. For them, it’s like a drill. Here we ourselves have a rest, eat watermelons and grapes, Marina takes pictures of us with a huge lens. We are waiting for the leaders. And, as it usually happens the first time, we skip it. Seryozha Donets and Vovan Levchenko are already running along the road from the BCP. I ran with them, chatted, got the news. Behind, very close, is Anton Golovin and one of the participants in the Russian Skyrunning Cup.I wonder what will happen at the finish line. Then I ran a little with the stadium, filmed the slopes of some of the participants. Then for this work I was praised by our video gurus, Vitalik and Dima. But this was not included in the video))) The whole team of photos and videos are real wizards, by the way.

And then we went to the finish line. Sasha is there. She works hard, but we managed to drink coffee together. After running to check the markings of the finishing piece, but on the approaches Anton is already rushing towards me. He is in a breakaway and wins today’s stage! Seryozha runs after him in a minute, and then in a minute Lyosha Vinogradov.With him and run to the finish line. Today, for Denis and me, the run is over, the second day is completely marked out by Topic with Sasha and Andrey with Ilyukha. And we even found time to wash and eat grapes in the Foros sanatorium.

In the same evening we met our foreign participants – the Belgians. They ran the Baikal Ice Marathon, and there someone told them about our race. And they, without thinking twice, said “f ** k yeah” and took tickets to … Kiev. From there they did not manage to get to the Crimea, then they flew to Moscow and from there they made it to the peninsula.These are truly desperate guys outside of politics. Respect and honor! They promised to tell all their acquaintances that Russians do not drink the blood of babies, as they say on TV.

Dinner and watching the video was a success, and then we were even able to get to the sanatorium named after Semashko and move into ward number 410. Yes, exactly ward. But what difference does it make if you have 4 hours of sleep left?

Chapter 4. Day Two.

Breakfast at 4.30. Begins. We are not in a hurry today, but the volume of work is simply wild: we set the 3rd day from the BPC to the top point and run away to the primary counting of the 4th day.According to conservative estimates – at least 45 km. Cross-trek in general. Fortunately, after the middle of the distance, in Schaslyve, a store is waiting for us. The walk turned out to be extremely pleasant, if not for one “but”: there is such a rule, if you do not take 5% of the CXR route, then there will be 90% of the difficulties of the entire distance. This rule worked 100%. After the ascent from Schastlivy Denis and I were expecting a first-class azimuth piece in the style of CXR16, only much longer. And it started: you mark every 5 meters, look for a path, find another option, shoot 100 meters of markings, re-mark.Meanwhile, the sun was already sinking. And we have not yet reached the yayla, there are still 10 kilometers ahead. Dinner and briefing are definitely without us, we are fighting for a few hours of sleep in bed. And only the matches in the backpack do not allow you to become completely sad. We frighten off the boar’s bed, we see a deer, we see off the sun, we find ourselves in a fog. So many events! It’s blowing on the yayla and it’s very cold – it’s time for an extreme sandwich and an extreme jacket. The culmination of the evening: what we took for a road turned out to be a trench under a gas pipe, and our distance runs for another kilometer along the ruthless off-road.

It starts to seem to me that we are already exactly in f * n. And that means it’s time to get out. Finally, the road. We mark it, since we are here. Descent. How will people run here? In the dark, he just looks insanely cool and dangerous. The steepness, multiplied by the slight discoordination from the 12-hour walk to that time, trips me up and I fly my back down the slope. Surprisingly, I land softly. From the consequences: a good abrasion and bruise on the left palm and, as it turned out later, a slight abrasion on the lower back.And I was saved by a huge cardboard sleeve from the marking tape, which folded softly and took the whole blow on itself. But, it would be time and honor to know. We run to the car, time is about 21.00. Even the faithful Suunto Spartan Ultra could not resist the last half hour, but they had a hard time – navigating the route takes a lot of effort. There is a holiday in the next ward today: Ilya’s birthday! And we are all about work. Tomorrow is a good day again, and there is not enough time to sleep. Sashechka helped as best she could, saved us dinner, and prepared everything for tomorrow’s departure from Semashko.There were about 3.5 hours of sleep left, which is definitely better than 2.5.

Chapter 5. Day Three.

Breakfasts earlier. Participants start today at 7 am. We check and spot the last descent of the third day, and then we move to “our azimuth”. Today everything happened again: a cool climb up Iograf, a meeting with Ilyukha and Sasha at the highest point of the third day, right in front of the leaders, from which, to the joy of Ilyukha, he even lost his phone (though later he found it thanks to the Track Back function in Suunto), and then a trip on a crazy ’76 car on a seemingly impassable mountain road.

On the plateau, meanwhile, it was not May, we ate a sandwich and decided to take a little break from each other: I ran to “our azimuth”, and Denis was already straight to the car, marking the descent of the fourth day along the way. Left alone, I realized that time was short. In order for Denis not to wait for me, I had to run properly. And here is a small gift of fate: I find the old road along the yayla, leading in the right direction. If all goes well, it will completely eliminate running along the stone trench. With every turn I am convinced that the road is good, the only question is when it will end, because it does not enter the forest in the right place.And one more small problem: now I am marking another path, and the previously marked one has yet to be filmed on the way back. We must take this into account.

As expected, the road led to the ridge in front of the forest. It’s already very good! Then I run along the markings, adding about as much as is already hanging. The main thing here is that the participant from one tape sees at least 3 more ahead of him. I go into the forest, again scare the wild boars on a large bed, trample and clean the bushes, dismantle debris from different branches.I even have a video!

I estimate the time and leave 2.5 hours for my return trip. I just manage to descend to the very beginning of the “azimuth”. I hang arrows, ribbons on trees, block unnecessary aisles. In short, almost a corridor made of tape. Now back. There is not much work on the way up, but it is there. I add another 20-25 percent. On the raid I take off the markings, once again I make my way through the stone trench and fall out onto the road. Now it’s time to go down. Well, let’s see how our fast participants see the markup.

I knock down, as my backpack allows me, but neatly. I thought it turned out quickly, but Strava says Vovan flew 4 minutes faster). The markings are visible, but not always, and if I ran at such a speed, not knowing the path, I would have had to be very nervous. It’s good at the bottom, we meet Denis, who overtook the car, could not turn on the reverse gear and threw it in the yard, and we go to pick Tyoma with Andrey to Gurzuf. Then, fortunately, to the Glade of Fairy Tales. Yes, yes, that very real Glade of Fairy Tales.

We take Sasha and go to the next Sanatorium, Pension or whatever was there at all.The head is no longer very clear, but there is a chance today even to wash before dinner. Rare luck! And now the most important thing: now we are not 300 people, but 500. The hall is simply unrealistically large, and also unrealistically full. I want to say how proud I am of everyone, but it turns out that it is not clear what – my thoughts are confused. It is impossible to get from point A to point B, on the way there is always an incredible number of friends and acquaintances. Once again I breathe in, the energy goes off scale!

We decided on the last day: I need to run away from the participants on the first ascent.Class, I start at 4 in the morning. So breakfast is at 3.30. And Sasha will run tomorrow, and even 50K. I must admit, I was worried, and very much. Her method of training without training sometimes scares me when it comes to serious distances. But at the same time I really needed her to come running to the finish line “through the front door.” Well, okay, she seems to be very serious herself. We equip it, fill the track, prepare in general.

Chapter 6. The fourth day.

I know! You just need to get up right away and not have time to get lazy as soon as the alarm clock rings.There won’t be a second chance. 3.45, breakfast. Two pancakes, porridge, another pancake. I can’t take it anymore – it’s too early. Dark Glade of Fairy Tales. “Hello, Den, come back, I forgot the flashlight in the car!” Without my faithful Petzl Nao, there will be nothing to do on the trail. Oh yes, the wind is still blowing and there are no stars in the sky. It looks like the weather has finally turned bad upstairs. I go out and think, well, now we are kapets. We slipped by for two years somehow, but this time it didn’t work.

Calling Sasha, I strongly recommend taking a warmer T-shirt.Promises to think. The higher, the more windy. I tie ribbons from the second and third branches to the first. Botkinskaya – Taraktash is generally an ambiguous place: continuous cuts of serpentine. I try to make it work without contradictions. And they are here at every step. I can already imagine how, after the finish, the participants call me for a conversation, wrap me in a marking tape and throw me into the sea. No, this will not happen! Better to be over-psyched than under-psyched!

And in the meantime, he pulls out the marking tape from his hands, part of the recently hung tape is lying on the path or nearby – the wind is raging.And the clock is ticking. There should have been a start already, and I was still under Taraktash. I think I will be in time if there are no surprises upstairs. Although how to say, without surprises, there is fog at the top, which means that the markup will have to be increased by three times. I am accelerating. Here comes the plateau! Flashlight – Oleg Chegodaev is already on duty. Warms up on me (photo – “Voooh there is Taraktash!”), And I run away further.

I meet the operator Slava in a down jacket, from which only the nose sticks out. Yes, there is wind and fog, but the temperature is above zero.I frighten Olya, who is filming a video from the jeep and does not expect to see anyone in the frame. The road is foggy, the markings are rare, as I feared. We have to increase it. But here there is no psychosis, because there is still nowhere to get out of this road. I run along the yayla for a long time, put the last arrows. Tick tock, tick tock.

It’s time to descend towards Schaslyve. Here I am sure that there will be little work, because Denis and I marked him normally, no one was supposed to shoot, and there is only one road. So I run, hanging almost nothing. Finally, there is a bond with Denis, who was walking towards me from Happy.Success, we made it! And besides, it was already quite warm, and as soon as we met, I took off my jacket. I am preparing for the meeting of the leaders, I want to shoot them on camera. But while I’m getting ready, none other than Zhenya Lepeshkin from Vladivostok is flying like a bullet! This is unexpected and very cool !!! I am very happy for this cool, powerful and positive guy! And then serious Kostya and Lyosha are running. They work hard, drive me away, ask me not to interfere. Well, okay, I’ll take pictures of others. After a few runners, Vovan puffs, I decide to run to the PP with him, but he can barely run – he hurt his back in the shower in the morning.But Vovan is a fighter to look for. We run chatting, gossiping gossip.

There is a paradise at the BCP: fruit and warmth. But we spent less than an hour there, then it’s time to go to the finish line, check the finishing piece and prepare our Important Business. On the way, we drop our volunteers from the PP to Yalta, it’s time for them to get on the plane. The guys got into such a turn for the first time and in complete, but seemingly happy, shock. In Gurzuf we meet with Tema and Andrey and go out and run upstairs. On the way, Kostya flies by – it seems already a three-time winner of CXR.And behind it there is a gap. The next one is Zhenya Lepeshkin! But Lyosha Vinogradov is close behind him, asks to lead him down, but I refuse, not out of harm, but only for reasonable competition between him and Zhenya.

We remain on duty on the lavender field with a beautiful view of Ayu-Dag, just at the last micro-ascent on the distance. We shout so that people start to run, we start a wave. I run downstairs with Sanka Ivakin, he literally has to be dragged to the cuts, his soul and legs already require asphalt. But we have a trail here, and we would have to run on the asphalt for another kilometer.

At the finish line, as usual, # has its own atmosphere. Taking a little breath of it, I go upstairs to the road where the car is parked. On the way I decide that I have to go to PP-3, catch Sasha. I rush with all my might, but on the spot it turns out that she has already escaped 15 minutes ago! Wow, it generally runs for 9-9.5 hours! Suddenly. It’s time for me to get ready. Rushing back, hanging out at the finish line. I warn Sanya and Pasha, our host and DJ, about my plan. They promise to play along. I’m even worried, I don’t seem to be able to say anything.Familiar people are finishing around, joy is everywhere: either from the beautiful mountains, or from the fact that everything is over. We are sorting out the current problems and of course we are waiting for Sasha.

And here she is running. And then I have nothing to write.

All this means a lot to me. To work like this, “survive” the race – apparently, this is what I need. This is our way of killing the routine, the only way a big deal is done. The unrealistic synergy of organizers and participants, this energy exchange, works wonders.And I hope that the numbers will converge, there will be partners and we will be able to continue this Big Bang. So I strongly recommend to all those who are not indifferent to hold their fingers for us and, if possible, assist) Well, we are moving – ahead of the IMB, two races in China, our trip to Italy, and then a small planned sports break under the quiet autumn whisper of Peter. I hope there will be something to tell on these pages covered with a small layer of dust.

Those who have read to the end can safely breathe out in the comments.

GERMES: X-CROSS 2017 | parkrun Zhukovsky

On the fourth of November in Bronnitsy near Moscow, the trail race GERMES: X-CROSS 2017 took place. By tradition, we ask residents of parkrun Zhukovsky to share their impressions of the start and tell how it was. Meet our heroes!

Alexey Tkachev, 47 km

“The main thing is that, despite the surface (frozen mud, muddy mud, ruts, and other delights), it is very similar to an asphalt marathon.And it doesn’t look like trail running in Crimea at all. Let’s just say it was a cross. Then everything is in place.

Good local organization. The track is marked, the locker rooms are warm, the gates, electronic markers, food, medals – everything is done, everything is without jambs (or I didn’t notice them). But after CXR, I already know that this is not enough, and much more can be done. Now about my race. I eat all the time. Every 5 km. Every 30 minutes. But then the food ends and I merge. I can’t understand if everyone who runs has the same addiction? I have not seen other participants devouring dried fruit gels with the same greed.

Distance of 47 km is two circles of 23.5 km each. According to my watch, it turned out 46. The first round – 2:08, I think it was an excellent result, and the second I lost, missed 3 places (slipped from 12th to 15th). And the time on the second lap is about 2:45. I often took a step, staggered, fell a couple of times. On the track in Strava you can see that I had a turning point, where I got up completely, took off my warm clothes, ate, went into the forest and then ran normally to the finish line, although I froze quickly.

There was one river, a meter and a half wide, a log was thrown across it, thin and wet.Someone fell there, but I was lucky, I slipped through. In general, there were no inevitable fords. ”

Andriy Shevchuk, 47 km

“The season turned out to be strange for me … without a major start. I tuned in for GRUT 50 in July, but got sick the day before. In the second half of the year, I ran several trails (and even 2 mountain ones) and an asphalt half marathon. But you need to somehow end the season by putting a bold point. There were two contenders in the neighborhood (both from Bronnitsy): the asphalt TITANIUM (50 km) and the GERMES: X-CROSS 2017 trail (47 km).Since this season I did not pay due attention to high-speed work and would hardly have shown everything that I could potentially do on asphalt, I decided to give my all on the trail.

This was my first ultra, albeit tiny. It was a chamber start, in terms of organization it was not chic, but everything you needed was there. The track is intelligible through forests and fields in 2 circles. The layout is perfect, it is impossible to get lost. There was a lot of dirt, which was frozen in places. I never run at random, I had certain estimates for the pace.Because the start is relatively flat, the pace can be planned. Based on past trail starts and training sessions, I was targeting an average pace of 5.30-6.00. As a result, it turned out like in a pharmacy – 5.29.

I started with one thought: this season there are only four and a half hours left to be patient. It was not easy to endure on the first lap the fact that you were bypassed by rivals one after the other, and you were running according to the plan. In addition, the start at 23.5 km and 47 km was common, and it was impossible to understand where you are approximately.When entering the second round, they told me that I was 8th (in fact, I was 10th), and I saw 3 opponents ahead, which I could really reach. I continued to work at a given pace and walked around them in turn after 5-7 km. There is no one ahead, motivation is falling, it is getting harder and harder to run. About 30 km far ahead I noticed another one, the volunteer suggested that there was about a minute between us. I am pursuing, but I can no longer add significantly, the distance does not decrease, hopes melt, pulls to slow down and take a break (no one is pressing me from behind).And suddenly I saw – the opponent stood in front of him, walking quietly, he was clearly feeling bad. You always have to endure to the end and fate will give you a chance! I rushed forward, caught up: the guy looked not at all good, gave him a drink of his own water and ran on. Already very close to the finish line, another rival loomed ahead, but there was not enough distance to fight, he ran 1 minute faster than me.

And I ended up 6th with a score of 4.12. And the main reward is that this year there is no need to run anywhere else. Tired of enduring, it’s time to rest and jerk off for pleasure. “

Marina Peretokina, 8.3 km

“Running for young people looking for thrills)) Well, my friend Olga and I from St. Petersburg – there too))) Cross-country cross with an incredible amount of fallen logs. Puddles under a thin layer of ice, into which the runners fell … The mud hiding under the foliage did not allow to relax. But the beauty of the forest made up for it all. The organization of the event is decent, the track is very well marked, the volunteers are friendly, ready to help and answer all questions.

Result 59:29. Only we were in our category W 50-54: I was 2nd, and Olga was 1st (46:29 – 4th overall (Elagin Ostrov club).

For me, after the parkcreen, this is the first race over 5 km. But nothing, endured, ran – the goal was achieved))) The support of friends at the finish line, who specially came to support Olga and me, gave strength at a distance. Thank you all very much!

Once again I was convinced: you have to pull yourself out of the favorable conditions by the collar towards adventures, because life is remembered only by this, and not by an empty pastime))) “.

Dmitry Kokhov covered 23.5 km with an excellent time of 2:21:01.

We congratulate all the participants of the race with another medal and another passed (and for some – the final) stage of the season! We are waiting for new victories!

the review and a case report

TOPICAL ISSUES IN CLINICAL TRANSPLANTOLOGY

TRANSPLANTOLOGY 1’2017 VOLUME 9 TRANSPLANTOLOGIYA 1’2017 vol. 9

TOPICAL ISSUES IN CLINICAL TRANSPLANTOLOGY

ACTUAL ISSUES OF TRANSPLANTATION

CLINICAL OBSERVATIONS

CASE REPORTS

2010; 139 (1): 130-139.e24. PMID: 20346360

DOI: 10.1053 / j.gastro.2010.03.044

16. Feng G., Rong H. The role of hemo-

dynamic and vasoactive substances on

hepatopulmonary syndrome. Eur Rev

Med Pharmacol Sci. 2014; 18 (3): 380–386.

PMID: 24563438

17. Fritz J.S., Fallon M.B., Kawut S.M.

Pulmonary vascular complications of

liver disease. Am J Respir Crit Care Med.

2013; 187 (2): 133-143. PMID: 23155142

DOI: 10.1164 / rccm. 201209-1583CI

18. Khan A.N., Al-Jahdali H., Abdullah K.,

et al. Pulmonary vascular complications

of chronic liver disease: Pathophy siology,

imaging, and treatment. Ann Thorac

Med. 2011; 6 (2): 57–65. PMID: 21572693

DOI: 10.4103 / 1817-1737.78412

19. Rodriguez-Roisin R., Krowka M.J.

Hepatopulmonary syndrome – a li ver-

induced lung vascular disorder. N

Engl J Med.2008; 358 (22): 2378-2387.

PMID: 18509123 DOI: 10.1056 / NEJM-

ra0707185

20. Macedo L.G., Lopes E.P. Hepato-

pulmonary syndrome: an update. Sao

Paulo Med J. 2009; 127 (4): 223-230.

PMID: 20011928

21. Polavarapu N., Tripathi D. Liver in

cardiopulmonary disease. Best Pract

Res Clin Gastroenterol. 2013; 27 (4): 497-

512. PMID: 24090938 DOI: 10.1016 / j.

bpg.2013.06.020

22.Chihara Y., Egawa H., Tsuboi T.,

et al. Immediate noninvasive ventila-

tion may improve mortality in patients

with hepatopulmonary syndrome after

liver transplantation. Liver Transpl.

2011; 17 (2): 144-148. PMID: 21280187

DOI: 10.1002 / lt. 22207

23. Zhang J., Yang W., Luo B., et al. The

role of CX (3) CL1 / CX (3) CR1 in pulmo-

nary angiogenesis and intravascular

monocyte accumulation in rat experimen-

tal hepatopulmonary syndrome.J Hepa-

tol. 2012; 57 (4): 752–758. PMID: 22659346

DOI: 10.1016 / j.jhep.2012.05.014

24. Zhang Z.J., Yang C.Q. Progress in

investigating the pathogenesis of hepa-

topulmonary syndrome. Hepatobiliary

Pancreat Dis Int. 2010; 9 (4): 355-360.

PMID: 20688597

25. Cremona G., Higenbottam T. W., May-

oral V., et al. Elevated exhaled nitric

oxide in patients with hepatopulmonary

syndrome.Eur Respir J. 1995; 8 (11): 1883-

1885. PMID: 8620957

26. Guo S.B., Duan Z.J., Li Q., Sun X.Y.

Effects of heme oxygenase-1 on pul-

monary function and structure in rats

with liver cirrhosis. Chin Med J (Engl).

2011; 124 (6): 918-922. PMID: 21518603

27. Ali O.M., Agarwal A., Akram S.

Platypnea orthodeoxia: a ‘laid-back’ case

of dyspnoea. BMJ Case Rep. 2013; 2013.

PII: bcr2012007810.PMID: 23362060

DOI: 10.1136 / bcr-2012-007810

28. Fallesen C.O., Sondergaard L., Nis-

sen H. Platypnoea-orthodeoxia: a rare

cause of severe dyspnoea. Ugeskr Lae-

ger. 2013; 175 (47A). PII: V11120636.

PMID: 25353093

29. Silverio Ade O., Guimarães D.C., Elias

L.F., et al. Are the spider angiomas skin

markers of hepatopulmonary syndrome?

Arq Gastroenterol. 2013; 50 (3): 175-179.

PMID: 24322187 DOI: 10.1590 / S0004-

28032013000200031

30. Gaber R., Ziada D.H., Kotb N.A., et

al. Detection of hepatopulmonary syn-

drome in patients with liver cirrhosis

using 3D contrast echocardiography.

Arab J Gastroenterol. 2012; 13 (1): 14–

19. PMID: 22560819 DOI: 10.1016 / j.

ajg. 2012.03.004.

31. Porres-Aguilar M., Gallegos-Orozco

J. F., Garcia H., et al.Pulmonary vascu-

lar complications in portal hyperten-

sion and liver disease: a concise review.

Rev Gastroenterol Mex. 2013; 78 (1): 35–

44. PMID: 23369639 DOI: 10.1016 / j.

rgmx.2012.10.004.

32. Awad Mel D., El-Arabi H.A., El-

Sharnouby K.A., Abo Dewan K.A. Diag-

nostic evaluation of hepatopulmonary

syndrome in Egyptian children with

chronic liver disease. J Egypt Soc Para-

sitol.2014; 44 (1): 97-112. PMID: 24961015

33. Kochar R., Tanikella R., Fallon

M.B. Serial pulse oximetry in hepa-

topulmonary syndrome. Dig Dis Sci.

2011; 56 (6): 1862-1868. PMID: 21327708

DOI: 10.1007 / s10620-011-1600-7

34. Suceveanu A.I., Mazilu L., Tomes-

cu D., et al. Screening of hepatopul-

monary syndrome (HPS) with CEUS

and pulse-oximetry in liver cirrhosis

patients eligible for liver transplant.

Chirurgia (Bucur). 2013; 108 (5): 684–688.

PMID: 24157113

35. Khabbaza J.E., Krasuski R.A., Tonelli

A.R. Intrapulmonary shunt confirmed

by intracardiac echocardiography in

the diagnosis of hepatopulmonary syn-

drome. Hepatology. 2013; 58 (4): 1514-1515.

PMID: 23696309 DOI: 10.1002 / hep. 26482

36. Fischer C.H., Campos O., Fernandes

W.B., et al. Role of contrast-enhanced

transesophageal echocardiography for

detection of and scoring intrapulmonary

vascular dilatation.Echocardiography.

2010; 27 (10): 1233-1237. PMID: 20584052

DOI: 10.1111 / j.1540-8175.2010.01228.x

37. Porres-Aguilar M., Altamirano

J.T., Torre-Delgadillo A., et al. Porto-

pulmonary hypertension and hepa-

topulmonary syndrome: a clinician-

oriented overview. Eur Respir Rev.

2012; 21 (125): 223-233. PMID: 22941887

DOI: 10.1183 / 09059180.00007211

38. El-Shabrawi M.H., Omran S., Wageeh

S., et al. (99m) Technetium-macroaggre-

gated albumin perfusion scan versus

contrast enhanced echocardiography in

the diagnosis of the hepatopulmonary

lung

syndrome in children with chronic liver

disease. Eur J. Gastroenterol Hepatol.

2010; 22 (8): 1006-1012. PMID: 20101183

DOI: 10.1097 / MEG.0b013e328336562e

39. Abrams G.A., Jaffe C.C., Hoffer

P.B., et al.Diagnostic utility of contrast

echocardiography and lung perfu-

sion scan in patients with hepatopul-

monary syndrome. Gastroenterology.

1995; 109 (4): 1283-1288. PMID: 7557096

40. Saad N.E., Lee D.E., Waldman D.L.,

Saad W.E. Pulmonary arterial coil embo-

lization for the management of persistent

type I hepatopulmonary syndrome after

liver transplantation. J Vasc Interv Radiol.

2007; 18 (12): 1576-1580.PMID: 18057294

DOI: 10.1016 / j.jvir.2007.08.008

41. Krowka M.J., Fallon M.B., Kawut

S.M., et al. International Liver Trans –

plant Society Practice Guidelines:

Diagnosis and Management of Hepa –

topulmonary Syndrome and Portopul –

monary Hypertension. Transplantation.

2016; 100 (7): 1440-1452. PMID: 27326810

DOI: 10.1097 / TP.0000000000001229

42. Gupta S., Faughnan M.E., Lilly L., et

al. Norfloxacin therapy for hepatopulmo-

nary syndrome: a pilot randomized con-

trolled trial. Clin Gastroenterol Hepatol.

2010; 8 (12): 1095-1098. PMID: 20816858

DOI: 10.1016 / j.cgh.2010.08.011

43. Kianifar H.R., Khalesi M., Mahmoodi

E., Afzal Aghaei M. Pentoxifylline in

hepatopulmonary syndrome. World J

Gastroenterol. 2012; 18 (35): 4912–4916.

PMID: 23002364 DOI: 10.3748 / wjg.v18.

i35.4912

44. Roma J., Balbi E., Pacheco-Moreira

L., et al. Methylene blue used as a bridge

to liver transplantation postoperative

recovery: a case report. Transplant Proc.

2010; 42 (2): 601-604. PMID: 20304203

DOI: 10.1016 / j.transproceed.2010.01.003

45. Chang C.C., Wang S.S., Hsieh H.G.