Smoking-Related Interstitial Lung Disease and Emphysema.

Diagnosis and treatment of patients with smoking-related lung diseases often requires multidisciplinary contributions to optimize care. Imaging plays a key role in characterizing the underlying disease, quantifying its severity, identifying potential complications, and directing management. The primary goal of this article is to provide an overview of the imaging findings and distinguishing features of smoking-related lung diseases, specifically, emphysema/chronic obstructive pulmonary disease, respiratory bronchiolitis-interstitial lung disease, smoking-related interstitial fibrosis, desquamative interstitial pneumonitis, combined pulmonary fibrosis and emphysema, pulmonary Langerhans cell histiocytosis, and E-cigarette or vaping related lung injury.

Practice patterns in reporting interstitial lung abnormality at a tertiary academic medical center.

PURPOSE: Interstitial lung abnormality (ILA) is a common finding on chest CTs and is associated with higher all-cause mortality. The 2020 Fleischner Society position paper standardized the terminology and definition of ILA. Despite these published guidelines, the extent to which radiologists use this term is unknown. We evaluated practice patterns for identification of ILAs among radiologists at a tertiary academic medical center. METHODS: In this retrospective review, we identified 157 radiology reports between January 1, 2010 through December 31, 2021 containing the phrase "interstitial lung abnormality" or "interstitial abnormality". After exclusions, 125 CT scans were reviewed by thoracic-trained radiologists using the sequential reading method. RESULTS: Seventy-seven (62%) patients were found to have ILA (69% subpleural fibrotic, 19% subpleural non-fibrotic, and 6% non-subpleural), nine (7%) were equivocal for ILA and 39 (31%) had no ILA. The term ILA was used exclusively by thoracic-trained radiologists except for two cases. Use of the term ILA has rapidly increased since the position paper publication (none from 2010-2017, one case in 2018, 20 cases in 2019, 41 cases in 2020, and 73 cases in 2021), and cases were typically very mild (1-25% of the lung). CONCLUSION: While there has been increased use of the term ILA among thoracic-trained radiologists, non-thoracic radiologists have essentially not begun to use the term. Almost one-third of cases labeled ILA on clinical reads were re-classified as not having ILA on research reads.

How Does Pruning Impact Long-Tailed Multi-Label Medical Image Classifiers?

Pruning has emerged as a powerful technique for compressing deep neural networks, reducing memory usage and inference time without significantly affecting overall performance. However, the nuanced ways in which pruning impacts model behavior are not well understood, particularly for long-tailed, multi-label datasets commonly found in clinical settings. This knowledge gap could have dangerous implications when deploying a pruned model for diagnosis, where unexpected model behavior could impact patient well-being. To fill this gap, we perform the first analysis of pruning's effect on neural networks trained to diagnose thorax diseases from chest X-rays (CXRs). On two large CXR datasets, we examine which diseases are most affected by pruning and characterize class "forgettability" based on disease frequency and co-occurrence behavior. Further, we identify individual CXRs where uncompressed and heavily pruned models disagree, known as pruning-identified exemplars (PIEs), and conduct a human reader study to evaluate their unifying qualities. We find that radiologists perceive PIEs as having more label noise, lower image quality, and higher diagnosis difficulty. This work represents a first step toward understanding the impact of pruning on model behavior in deep long-tailed, multi-label medical image classification. All code, model weights, and data access instructions can be found at https://github.com/VITA-Group/PruneCXR.

How Does Pruning Impact Long-Tailed Multi-label Medical Image Classifiers?

Pruning has emerged as a powerful technique for compressing deep neural networks, reducing memory usage and inference time without significantly affecting overall performance. However, the nuanced ways in which pruning impacts model behavior are not well understood, particularly for long-tailed, multi-label datasets commonly found in clinical settings. This knowledge gap could have dangerous implications when deploying a pruned model for diagnosis, where unexpected model behavior could impact patient well-being. To fill this gap, we perform the first analysis of pruning's effect on neural networks trained to diagnose thorax diseases from chest X-rays (CXRs). On two large CXR datasets, we examine which diseases are most affected by pruning and characterize class "forgettability" based on disease frequency and co-occurrence behavior. Further, we identify individual CXRs where uncompressed and heavily pruned models disagree, known as pruning-identified exemplars (PIEs), and conduct a human reader study to evaluate their unifying qualities. We find that radiologists perceive PIEs as having more label noise, lower image quality, and higher diagnosis difficulty. This work represents a first step toward understanding the impact of pruning on model behavior in deep long-tailed, multi-label medical image classification. All code, model weights, and data access instructions can be found at https://github.com/VITA-Group/PruneCXR.

Screening Chest CT Prior to Allogenic Hematopoietic Stem Cell Transplantation.

Pulmonary complications constitute a major cause of morbidity and mortality in the post-allogenic hematopoietic stem cell transplantation (alloHSCT) period. Although chest X-ray (CXR) is customarily used for screening, we have used chest computed tomography (CT) scans. To characterize the prevalence of abnormalities and explore their impact on alloHSCT eligibility and outcomes post-transplantation, we conducted a retrospective analysis using real-world data collected at our center for adult patients who were evaluated for alloHSCT between January 2013 and December 2020 and identified 511 eligible patients. The most common primary disease was acute myeloid leukemia, in 49% of patients, followed by myelodysplastic syndrome (23%), lymphoma (11%), and acute lymphocytic leukemia (10%). Abnormal screening chest CT results were found in 199 patients (39%). The most frequent detected abnormality was pulmonary nodule, in 78 patients (35%), followed by consolidation in 42 (19%), ground-glass opacification in 33 (15%), bronchitis and bronchiolitis in 25 (11%), pleural effusions in 14 (6%), and new primary cancer in 7 (2%). CXR detected abnormalities in only approximately one-half of the patients (48%) with an abnormal chest CT scan. Among the 199 patients with an abnormal chest CT scan, 98 (49%) underwent further assessment and/or intervention before transplantation. The most common workup was pulmonary consultation in 32%, followed by infectious diseases consultation in 24%. Lung biopsy was obtained in 20%, and antimicrobial therapy was initiated after confirming an infection diagnosis in 20%. Patients with an abnormal chest CT scan demonstrated worse overall survival (P = .032), nonrelapse mortality (P = .015), and pulmonary-related mortality (P < .001) compared to those with a normal chest CT scan. Our study suggests that pretransplantation screening chest CT is beneficial in uncovering invasive infections and underlying malignancies and allows for appropriate interventions before alloHSCT to prevent potentially serious post-transplantation complications without causing a delay in alloHSCT. Nevertheless, abnormal CT findings prior to transplantation may be associated with overall worse prognosis.

Deciphering the origin and therapeutic targets of cancer of unknown primary: a case report that illustrates the power of integrative whole-exome and transcriptome sequencing analysis.

Cancer of unknown primary (CUP) represents a significant diagnostic and therapeutic challenge, being the third to fourth leading cause of cancer death, despite advances in diagnostic tools. This article presents a successful approach using a novel genomic analysis in the evaluation and treatment of a CUP patient, leveraging whole-exome sequencing (WES) and RNA sequencing (RNA-seq). The patient, with a history of multiple primary tumors including urothelial cancer, exhibited a history of rapid progression on empirical chemotherapy. The application of our approach identified a molecular target, characterized the tumor expression profile and the tumor microenvironment, and analyzed the origin of the tumor, leading to a tailored treatment. This resulted in a substantial radiological response across all metastatic sites and the predicted primary site of the tumor. We argue that a comprehensive genomic and molecular profiling approach, like the BostonGene© Tumor Portrait, can provide a more definitive, personalized treatment strategy, overcoming the limitations of current predictive assays. This approach offers a potential solution to an unmet clinical need for a standardized approach in identifying the tumor origin for the effective management of CUP.

Incidence, clinical associations and outcomes of intrathoracic complications with and without ARDS in COVID-19 pneumonia.

PURPOSE: To determine the incidence and clinical predictors of intrathoracic complications in COVID-19 patients, and the association with outcomes. METHODS: In this retrospective cross-sectional study, we included 976 patients (age 61 ± 17 years, 62% male) who tested positive for SARS-CoV-2 between March 3-April 4, 2020 and underwent chest imaging. 3836 radiographs from 976 patients and 105 CTs from 88 patients were reviewed for intrathoracic complications, including pneumothorax, pneumomediastinum, pneumopericardium, lobar collapse, pleural effusion, and pneumatocele formation. RESULTS: There was a high rate of intrathoracic complications (197/976, 20%). Pleural effusion was the most common complication (168/976, 17%). Pneumothorax (30/976, 3%) and pneumatoceles (9/88, 10%) were also frequent. History of hypertension and high initial CXR severity score were independent risk factors for complications. Patients with any intrathoracic complication during admission had an over 11-fold risk of ICU admission (adjusted odds ratio [aOR] 11.2, p < 0.0001) and intubation (aOR 12.4, p < 0.0001), over 50% reduction in successful extubation (aOR 0.49, p = 0.02) and longer length of stay (median 13 versus 5 days, p < 0.0001). There was no difference in overall survival between patients with and without any complication (log-rank p = 0.94). CONCLUSION: In COVID-19 patients who underwent chest imaging, 1 in 5 patients have an intrathoracic complication, which are associated with higher level of care and prolonged hospital stay. Hypertension history and high CXR severity score confer an increased risk of complication. SUMMARY: Intrathoracic complications in COVID-19 are common and are predictive of ICU admission, need for intubation, less successful extubation, and longer length of stay but are not predictive of mortality.

Management of Incidental Pulmonary Nodules: Influencing Patient Care Through Subspecialized Imaging Review.

OBJECTIVE: To evaluate whether thoracic radiologist review of computed tomography-detected incidental pulmonary nodules initially reported by non-thoracic imagers would change management recommendations. MATERIALS AND METHODS: The Radiology Consultation Service identified 468 computed tomography scans (one per patient) performed through the adult emergency department from August 2018 through December 2020 that mentioned the presence of a pulmonary nodule. Forty percent (186/468) were read by thoracic radiologists and 60% (282/468) were read by non-thoracic radiologists. The Radiology Consultation Service contacted all patients in order to assess risk factors for lung malignancy. Sixty-seven patients were excluded because they were unreachable, declined participation, or were actively followed by a pulmonologist or oncologist. A thoracic radiologist assessed the nodule and follow up recommendations in all remaining cases. RESULTS: A total of 215 cases were re-reviewed by thoracic radiologists. The thoracic radiologist disagreed with the initial nodule recommendations in 38% (82/215) of cases and agreed in 62% (133/215) of cases. All discordant cases resulted in a change in management by the thoracic radiologist with approximately one-third (33%, 27/82) decreasing imaging utilization and two-thirds (67%, 55/82) increasing imaging utilization. Nodules were deemed benign and follow up eliminated in 11% (9/82) of discordant cases. DISCUSSION: Our study illustrates that nodule review by thoracic radiologists results in a change in management in a large percentage of patients. Continued research is needed to determine whether subspecialty imaging review results in increased or more timely lung cancer detection.

Central paradiaphragmatic middle lobe involvement in nonspecific interstitial pneumonia.

OBJECTIVES: Nonspecific interstitial pneumonia (NSIP) lacks specific diagnostic guidelines or criteria for imaging diagnosis, and the need for more reliable computed tomography (CT) characterization remains. We hypothesized that central paradiaphragmatic middle lobe (ML) involvement is present in most patients with NSIP. The purpose of this study was to evaluate the prevalence of ML involvement and thus to assess its potential as a unique feature of NSIP. METHODS: We conducted a retrospective CT-imaging review of 40 patients with biopsy-proven (7/40, 18%) or clinically established (33/40, 82%) NSIP. Three subspecialty-trained thoracic radiologists reviewed CTs for ML involvement both independently and in consensus, and additional CT findings previously described in NSIP independently. RESULTS: ML involvement was present in most cases (70%, 28/40, independent review, 78%, 31/40, consensus reading), with substantial agreement among all three readers (κ = 0.65). Fibrosis was present in almost all cases (93%, 37/40). Subpleural sparing occurred in one-third of patients (30%, 12/40). Homogeneity (48%, 19/40), central bronchiectasis (45%, 18/40), and peripheral bronchiectasis (53%, 21/40) were present in about half of patients. Apart from substantial inter-reader agreement on fibrosis (κ = 0.65), the above-mentioned imaging characteristics had fair to slight universal agreement (κ = 0.07-0.30). CONCLUSIONS: Central paradiaphragmatic ML ground glass attenuation superimposed on reticulation and traction bronchiectasis occurs in most patients with NSIP, with high interobserver agreement. KEY POINTS: • Central paradiaphragmatic middle lobe ground glass attenuation superimposed on reticulation and traction bronchiectasis is common in nonspecific interstitial pneumonia (NSIP). • This finding occurs more frequently than subpleural sparing and has a better interobserver agreement.

Thoracic Manifestations of Rheumatoid Arthritis.

Rheumatoid arthritis (RA) is one of the most common chronic systemic inflammatory diseases and the most common chronic inflammatory arthritis. Classically a progressive symmetric polyarthritis, RA is characterized by inflammation, erosions, bone loss, and joint destruction. Up to half of patients with RA exhibit extra-articular manifestations (EAMs), which may precede articular disease and are more common in patients with seropositive RA (patients with detectable serum levels of rheumatoid factor and/or anticitrullinated peptide antibodies). Cardiovascular and pulmonary EAMs are the largest contributors to morbidity and mortality in RA and may be especially devastating. Imaging has a significant role in diagnosing these EAMs and assessing response to treatment. Although treatment with disease-modifying antirheumatic drugs has redefined the natural history of RA and helped many patients achieve low disease activity, patients are at risk for treatment-related complications, as well as infections. The clinical features of drug-induced lung disease and infection can overlap considerably with those of EAMs, presenting a diagnostic challenge. Radiologists, by recognizing the imaging characteristics and evolution of these various processes, are essential in diagnosing and distinguishing among EAMs, treatment-related complications, and unrelated processes and formulating an appropriate differential diagnosis. Moreover, recognizing these disease processes at imaging and contextualizing imaging findings with clinical information and laboratory and pathologic findings can facilitate definitive diagnosis and proper treatment. The authors review the articular and extra-articular thoracic imaging manifestations of RA, including cardiovascular, respiratory, and pleural diseases, as well as treatment-related complications and common infections. ©RSNA, 2021.

Myocardial Infarction With Nonobstructive Coronary Arteries (MINOCA): Potential Etiologies, Mimics and Imaging Findings.

Myocardial infarction with nonobstructive coronary arteries (MINOCA) occurs when a patient presents with positive cardiac enzymes in the absence of obstructive atherosclerosis on coronary angiography. Several hypotheses for the pathogenesis of MINOCA have been suggested and multiple potential underlying etiologies have been reported. This review will outline the reported causes of MINOCA and associated major imaging features. In doing so, it will increase awareness of this entity and equip cardiac imagers with the knowledge to appropriately tailor imaging to make a prompt and accurate diagnosis.

A Case-Based Review of Vaping-Induced Injury-Pulmonary Toxicity and Beyond.

The last 10 years has seen a steady rise in the use of electronic cigarettes ("e-cigarettes" or ECIGs) or "vape pens." Though initially developed to assist with smoking cessation, use among adolescents has been particularly high. A concomitant rise in ECIG-related injuries disproportionately affecting young patients has been recognized. This unique case series highlights both pulmonary and extra-pulmonary ECIG-induced injuries including vape tip ingestion, maxillofacial fractures after vape pen explosion, myocarditis, and several different manifestations of vaping-associated lung injury. Becoming familiar with expected imaging findings in the wide array of ECIG-induced complications will help radiologists recognize these findings, recommend further imaging as needed, facilitate early diagnosis by help referring clinicians elicit the relevant history from patients, and expedite appropriate treatment.

Imaging Utilization and Outcomes in Vulnerable Populations during COVID-19 in New York City.

BACKGROUND: Coronavirus disease 2019 (COVID-19) affects vulnerable populations (VP) adversely. PURPOSE: To evaluate overall imaging utilization in vulnerable subgroups (elderly, racial/ethnic minorities, socioeconomic status [SES] disadvantage) and determine if a particular subgroup has worse outcomes from COVID-19. MATERIALS/METHODS: Of 4110 patients who underwent COVID-19 testing from March 3-April 4, 2020 at NewYork-Presbyterian Hospital (NYP) health system, we included 1121 COVID-19 positive adults (mean age 59±18 years, 59% male) from two academic hospitals and evaluated imaging utilization rates and outcomes, including mortality. RESULTS: Of 897 (80%) VP, there were 465 (41%) elderly, 380 (34%) racial/ethnic minorities, and 479 (43%) SES disadvantage patients. Imaging was performed in 88% of patients and mostly portable/bedside studies, with 87% of patients receiving chest radiographs. There were 83% hospital admissions, 25% ICU admissions, 23% intubations, and 13% deaths. Elderly patients had greater imaging utilization, hospitalizations, ICU/intubation requirement, longer hospital stays, and >4-fold increase in mortality compared to non-elderlies (adjusted hazard ratio[aHR] 4.79, p<0.001). Self-reported minorities had fewer ICU admissions (p=0.03) and reduced hazard for mortality (aHR 0.53, p=0.004; complete case analysis: aHR 0.39, p<0.001 excluding "not reported"; sensitivity analysis: aHR 0.61, p=0.005 "not reported" classified as minorities) with similar imaging utilization, compared to non-minorities. SES disadvantage patients had similar imaging utilization and outcomes as compared to their counterparts. CONCLUSIONS: In a predominantly hospitalized New York City cohort, elderly patients are at highest mortality risk. Racial/ethnic minorities and SES disadvantage patients fare better or similarly to their counterparts, highlighting the critical role of access to inpatient medical care during the COVID-19 pandemic.

Isolated Cystic Lung Disease: An Algorithmic Approach to Distinguishing Birt-Hogg-Dubé Syndrome, Lymphangioleiomyomatosis, and Lymphocytic Interstitial Pneumonia.

OBJECTIVE. Birt-Hogg-Dubé (BHD) syndrome, lymphangioleiomyomatosis (LAM), and lymphocytic interstitial pneumonia (LIP) frequently present as isolated cystic lung disease and can be challenging to distinguish. If imaging findings are otherwise unremarkable, the radiologist is unaided by ancillary CT findings in narrowing the diagnosis. We hypothesized that the distribution and morphologic features of lung cysts could be used to differentiate BHD syndrome, LAM, and LIP. Therefore, the purpose of this study was to characterize the CT appearances of these conditions and create a practical CT-based algorithm to differentiate among them. MATERIALS AND METHODS. The study was a retrospective review of the CT images of 16 patients with BHD syndrome, 17 patients with LAM, and 14 patients with LIP. On the basis of the data collected, a CT-based algorithm was created, and the CT images were reviewed again. RESULTS. Lower lung-predominant cysts were significantly more likely to be found in patients with BHD syndrome (100% of patients) or LIP (71-93% of patients) than in patients with LAM (6-12% of patients), who were more likely to have diffuse cysts. Compared with patients with LIP or LAM, patients with BHD syndrome were significantly more likely to have elliptical (floppy) paramediastinal cysts (88-94% of patients with BHD syndrome, 36-43% of patients with LIP, and 6-12% of patients with LAM) or a disproportionate number of paramediastinal cysts (69-88% of patients with BHD syndrome, 0-14% of patients with LIP, and 0-6% of patients with LAM). Our algorithm enabled differentiation of BHD syndrome, LAM, and LIP with a high level of accuracy and high interreader agreement (κ = 0.809). CONCLUSION. Radiologists can use the proposed CT-based algorithm to prospectively and confidently suggest one of these disorders as the favored diagnosis. Of importance, this will allow diagnosing the disorder early and accurately, screening for comorbidities, and prevention of potential complications.

Congenital anomalies of the pulmonary arteries: an imaging overview.

Congenital pulmonary artery anomalies represent a diverse group of abnormalities, ranging from asymptomatic incidental findings to causes of sudden cardiac death. While some may be recognized in childhood, others may be found incidentally in adulthood. We review the clinical and imaging findings in patients with congenital anomalies of the pulmonary arteries, including valvular and perivavular anomalies as well as abnormal narrowing, course and communications of the pulmonary arteries. We also discuss the role of various imaging modalities in the evaluation of these patients. It is vital to be aware of the key radiologic manifestations and associated haemodynamic consequences in these conditions in order to facilitate accurate diagnosis and prognostic stratification.

Malignant Pleural Mesothelioma: Are There Imaging Characteristics Associated With Different Histologic Subtypes on Computed Tomography?

OBJECTIVES: Determine imaging characteristics specific to epithelioid (eMPM), sarcomatoid (sMPM), and biphasic (bMPM) subtypes of malignant pleural mesothelioma (MPM) on computed tomography. METHODS: Preoperative computed tomography scans of patients with MPM were retrospectively assessed for numerous features including primary affected side, volume loss, pleural thickness, pleural calcifications, pleural effusion, and lymphadenopathy. RESULTS: One hundred twenty-five patients with MPM were included. Histologic subdivision was 97 eMPM (77%), 17 bMPM (14%), and 11 sMPM (9%). Nonepithelioid MPM (bMPM and sMPM) was more likely than eMPM to have calcified pleural plaques (P = 0.035). Analyzed separately, bMPM and sMPM each demonstrated calcified plaques more frequently than eMPM, and sMPM more often had internal mammary nodes; however, P values did not reach significance (P = 0.075 and 0.071, respectively). CONCLUSIONS: Calcified plaques are significantly more common in nonepithelioid subtypes compared with eMPM. Given the different prognoses and management of MPM subtypes, accurate noninvasive subtype classification is clinically vital.