From Pixels to Prognosis: Unlocking the Potential of Deep Learning in Fibrotic Lung Disease Imaging Analysis.

The licensing of antifibrotic therapy for fibrotic lung diseases, including idiopathic pulmonary fibrosis IPF has created an urgent need for reliable biomarkers to predict disease progression and treatment response. Some patients experience stable disease trajectories, while others deteriorate rapidly, making treatment decisions challenging. High-resolution chest CT has become crucial for diagnosis, but visual assessments by radiologists suffer from low reproducibility and high interobserver variability. To address these issues, computer-based image analysis, called quantitative CT, has emerged. However, many quantitative CT methods rely on human input for training, therefore potentially incorporating human error into computer training. Rapid advances in artificial intelligence, specifically deep learning, aim to overcome this limitation by enabling autonomous quantitative analysis. While promising, deep learning also presents challenges including the need to minimize algorithm biases, ensuring explainability, and addressing accessibility and ethical concerns. This review explores the development and application of deep learning in improving the imaging process for fibrotic lung disease.

Progressive Pulmonary Fibrosis and Interstitial Lung Abnormalities: AJR Expert Panel Narrative Review.

Progressive pulmonary fibrosis (PPF) and interstitial lung abnormalities (ILA) are relatively new concepts in interstitial lung disease (ILD) imaging and clinical management. Recognition of signs of PPF, as well as identification and classification of ILA, are important tasks during chest high-resolution CT interpretation, to optimize management of patients with ILD and those at risk of developing ILD. However, following professional society guidance, the role of imaging surveillance remains unclear in stable patients with ILD, asymptomatic patients with ILA who are at risk of progression, and asymptomatic patients at risk of developing ILD without imaging abnormalities. In this AJR Expert Panel Narrative Review, we summarize the current knowledge regarding PPF and ILA and describe the range of clinical practice with respect to imaging patients with ILD, those with ILA, and those at risk of developing ILD. In addition, we offer suggestions to help guide surveillance imaging in areas with an absence of published guidelines, where such decisions are currently driven primarily by local pulmonologists' preference.

Towards the adoption of quantitative computed tomography in the management of interstitial lung disease.

The shortcomings of qualitative visual assessment have led to the development of computer-based tools to characterise and quantify disease on high-resolution computed tomography (HRCT) in patients with interstitial lung diseases (ILDs). Quantitative CT (QCT) software enables quantification of patterns on HRCT with results that are objective, reproducible, sensitive to change and predictive of disease progression. Applications developed to provide a diagnosis or pattern classification are mainly based on artificial intelligence. Deep learning, which identifies patterns in high-dimensional data and maps them to segmentations or outcomes, can be used to identify the imaging patterns that most accurately predict disease progression. Optimisation of QCT software will require the implementation of protocol standards to generate data of sufficient quality for use in computerised applications and the identification of diagnostic, imaging and physiological features that are robustly associated with mortality for use as anchors in the development of algorithms. Consortia such as the Open Source Imaging Consortium have a key role to play in the collation of imaging and clinical data that can be used to identify digital imaging biomarkers that inform diagnosis, prognosis and response to therapy.

Zinpentraxin Alfa for Idiopathic Pulmonary Fibrosis: The Randomized Phase III STARSCAPE Trial.

Rationale: A phase II trial reported clinical benefit over 28 weeks in patients with idiopathic pulmonary fibrosis (IPF) who received zinpentraxin alfa. Objectives: To investigate the efficacy and safety of zinpentraxin alfa in patients with IPF in a phase III trial. Methods: This 52-week phase III, double-blind, placebo-controlled, pivotal trial was conducted at 275 sites in 29 countries. Patients with IPF were randomized 1:1 to intravenous placebo or zinpentraxin alfa 10 mg/kg every 4 weeks. The primary endpoint was absolute change from baseline to Week 52 in FVC. Secondary endpoints included absolute change from baseline to Week 52 in percent predicted FVC and 6-minute walk distance. Safety was monitored via adverse events. Post hoc analysis of the phase II and phase III data explored changes in FVC and their impact on the efficacy results. Measurements and Main Results: Of 664 randomized patients, 333 were assigned to placebo and 331 to zinpentraxin alfa. Four of the 664 randomized patients were never administered study drug. The trial was terminated early after a prespecified futility analysis that demonstrated no treatment benefit of zinpentraxin alfa over placebo. In the final analysis, absolute change from baseline to Week 52 in FVC was similar between placebo and zinpentraxin alfa (-214.89 ml and -235.72 ml; P = 0.5420); there were no apparent treatment effects on secondary endpoints. Overall, 72.3% and 74.6% of patients receiving placebo and zinpentraxin alfa, respectively, experienced one or more adverse events. Post hoc analysis revealed that extreme FVC decline in two placebo-treated patients resulted in the clinical benefit of zinpentraxin alfa reported by phase II. Conclusions: Zinpentraxin alfa treatment did not benefit patients with IPF over placebo. Learnings from this program may help improve decision making around trials in IPF. Clinical trial registered with www.clinicaltrials.gov (NCT04552899).

High-resolution CT phenotypes in pulmonary sarcoidosis: a multinational Delphi consensus study.

One view of sarcoidosis is that the term covers many different diseases. However, no classification framework exists for the future exploration of pathogenetic pathways, genetic or trigger predilections, patterns of lung function impairment, or treatment separations, or for the development of diagnostic algorithms or relevant outcome measures. We aimed to establish agreement on high-resolution CT (HRCT) phenotypic separations in sarcoidosis to anchor future CT research through a multinational two-round Delphi consensus process. Delphi participants included members of the Fleischner Society and the World Association of Sarcoidosis and other Granulomatous Disorders, as well as members' nominees. 146 individuals (98 chest physicians, 48 thoracic radiologists) from 28 countries took part, 144 of whom completed both Delphi rounds. After rating of 35 Delphi statements on a five-point Likert scale, consensus was achieved for 22 (63%) statements. There was 97% agreement on the existence of distinct HRCT phenotypes, with seven HRCT phenotypes that were categorised by participants as non-fibrotic or likely to be fibrotic. The international consensus reached in this Delphi exercise justifies the formulation of a CT classification as a basis for the possible definition of separate diseases. Further refinement of phenotypes with rapidly achievable CT studies is now needed to underpin the development of a formal classification of sarcoidosis.

Imaging Features of Autoimmune Disease-Related Interstitial Lung Diseases.

Interstitial lung diseases (ILDs) associated with autoimmune diseases show characteristic signs of imaging. Radiologic signs are also used in the identification of ILDs with features suggestive of autoimmune disease that do not meet the criteria for a specific autoimmune disease. Radiologists play a key role in identifying these signs and assessing their relevance as part of multidisciplinary team discussions. A radiologist may be the first health care professional to pick up signs of autoimmune disease in a patient referred for assessment of ILD or with suspicion for ILD. Multidisciplinary team discussion of imaging findings observed during follow-up may inform a change in diagnosis or identify progression, with implications for a patient's treatment regimen. This article describes the imaging features of autoimmune disease-related ILDs and the role of radiologists in assessing their relevance.

Combination of BAL and Computed Tomography Differentiates Progressive and Non-progressive Fibrotic Lung Diseases.

Rationale: Identifying patients with pulmonary fibrosis (PF) at risk of progression can guide management. Objectives: To explore the utility of combining baseline BAL and computed tomography (CT) in differentiating progressive and nonprogressive PF. Methods: The derivation cohort consisted of incident cases of PF for which BAL was performed as part of a diagnostic workup. A validation cohort was prospectively recruited with identical inclusion criteria. Baseline thoracic CT scans were scored for the extent of fibrosis and usual interstitial pneumonia (UIP) pattern. The BAL lymphocyte proportion was recorded. Annualized FVC decrease of >10% or death within 1 year was used to define disease progression. Multivariable logistic regression identified the determinants of the outcome. The optimum binary thresholds (maximal Wilcoxon rank statistic) at which the extent of fibrosis on CT and the BAL lymphocyte proportion could distinguish disease progression were identified. Measurements and Main Results: BAL lymphocyte proportion, UIP pattern, and fibrosis extent were significantly and independently associated with disease progression in the derivation cohort (n = 240). Binary thresholds for increased BAL lymphocyte proportion and extensive fibrosis were identified as 25% and 20%, respectively. An increased BAL lymphocyte proportion was rare in patients with a UIP pattern (8 of 135; 5.9%) or with extensive fibrosis (7 of 144; 4.9%). In the validation cohort (n = 290), an increased BAL lymphocyte proportion was associated with a significantly lower probability of disease progression in patients with nonextensive fibrosis or a non-UIP pattern. Conclusions: BAL lymphocytosis is rare in patients with extensive fibrosis or a UIP pattern on CT. In patients without a UIP pattern or with limited fibrosis, a BAL lymphocyte proportion of ⩾25% was associated with a lower likelihood of progression.

Exploring computer-based imaging analysis in interstitial lung disease: opportunities and challenges.

The advent of quantitative computed tomography (QCT) and artificial intelligence (AI) using high-resolution computed tomography data has revolutionised the way interstitial diseases are studied. These quantitative methods provide more accurate and precise results compared to prior semiquantitative methods, which were limited by human error such as interobserver disagreement or low reproducibility. The integration of QCT and AI and the development of digital biomarkers has facilitated not only diagnosis but also prognostication and prediction of disease behaviour, not just in idiopathic pulmonary fibrosis in which they were initially studied, but also in other fibrotic lung diseases. These tools provide reproducible, objective prognostic information which may facilitate clinical decision-making. However, despite the benefits of QCT and AI, there are still obstacles that need to be addressed. Important issues include optimal data management, data sharing and maintenance of data privacy. In addition, the development of explainable AI will be essential to develop trust within the medical community and facilitate implementation in routine clinical practice.

Systemic sclerosis associated interstitial lung disease: a conceptual framework for subclinical, clinical and progressive disease.

OBJECTIVES: To establish a framework by which experts define disease subsets in systemic sclerosis associated interstitial lung disease (SSc-ILD). METHODS: A conceptual framework for subclinical, clinical and progressive ILD was provided to 83 experts, asking them to use the framework and classify actual SSc-ILD patients. Each patient profile was designed to be classified by at least four experts in terms of severity and risk of progression at baseline; progression was based on 1-year follow-up data. A consensus was reached if ≥75% of experts agreed. Experts provided information on which items were important in determining classification. RESULTS: Forty-four experts (53%) completed the survey. Consensus was achieved on the dimensions of severity (75%, 60 of 80 profiles), risk of progression (71%, 57 of 80 profiles) and progressive ILD (60%, 24 of 40 profiles). For profiles achieving consensus, most were classified as clinical ILD (92%), low risk (54%) and stable (71%). Severity and disease progression overlapped in terms of framework items that were most influential in classifying patients (forced vital capacity, extent of lung involvement on high resolution chest CT [HRCT]); risk of progression was influenced primarily by disease duration. CONCLUSIONS: Using our proposed conceptual framework, international experts were able to achieve a consensus on classifying SSc-ILD patients along the dimensions of disease severity, risk of progression and progression over time. Experts rely on similar items when classifying disease severity and progression: a combination of spirometry and gas exchange and quantitative HRCT.

Quantitative computed tomography and machine learning: recent data in fibrotic interstitial lung disease and potential role in pulmonary sarcoidosis.

PURPOSE OF REVIEW: The aim of this study was to summarize quantitative computed tomography (CT) and machine learning data in fibrotic lung disease and to explore the potential application of these technologies in pulmonary sarcoidosis. RECENT FINDINGS: Recent data in the use of quantitative CT in fibrotic interstitial lung disease (ILD) are covered. Machine learning includes deep learning, a branch of machine learning particularly suited to medical imaging analysis. Deep learning imaging biomarker research in ILD is currently undergoing accelerated development, driven by technological advances in image processing and analysis. Fundamental concepts and goals related to deep learning imaging research in ILD are discussed. Recent work highlighted in this review has been performed in patients with idiopathic pulmonary fibrosis (IPF). Quantitative CT and deep learning have not been applied to pulmonary sarcoidosis, although there are recent deep learning data in cardiac sarcoidosis. SUMMARY: Pulmonary sarcoidosis presents unsolved problems for which quantitative CT and deep learning may provide unique solutions: in particular, the exploration of the long-standing question of whether sarcoidosis should be viewed as a single disease or as an umbrella term for disorders that might usefully be considered as separate diseases.

Quantitative computed tomography predicts outcomes in idiopathic pulmonary fibrosis.

BACKGROUND AND OBJECTIVE: Prediction of disease course in patients with progressive pulmonary fibrosis remains challenging. The purpose of this study was to assess the prognostic value of lung fibrosis extent quantified at computed tomography (CT) using data-driven texture analysis (DTA) in a large cohort of well-characterized patients with idiopathic pulmonary fibrosis (IPF) enrolled in a national registry. METHODS: This retrospective analysis included participants in the Australian IPF Registry with available CT between 2007 and 2016. CT scans were analysed using the DTA method to quantify the extent of lung fibrosis. Demographics, longitudinal pulmonary function and quantitative CT metrics were compared using descriptive statistics. Linear mixed models, and Cox analyses adjusted for age, gender, BMI, smoking history and treatment with anti-fibrotics were performed to assess the relationships between baseline DTA, pulmonary function metrics and outcomes. RESULTS: CT scans of 393 participants were analysed, 221 of which had available pulmonary function testing obtained within 90 days of CT. Linear mixed-effect modelling showed that baseline DTA score was significantly associated with annual rate of decline in forced vital capacity and diffusing capacity of carbon monoxide. In multivariable Cox proportional hazard models, greater extent of lung fibrosis was associated with poorer transplant-free survival (hazard ratio [HR] 1.20, p < 0.0001) and progression-free survival (HR 1.14, p < 0.0001). CONCLUSION: In a multi-centre observational registry of patients with IPF, the extent of fibrotic abnormality on baseline CT quantified using DTA is associated with outcomes independent of pulmonary function.

Deep Learning-based Outcome Prediction in Progressive Fibrotic Lung Disease Using High-Resolution Computed Tomography.

Rationale: Reliable outcome prediction in patients with fibrotic lung disease using baseline high-resolution computed tomography (HRCT) data remains challenging. Objectives: To evaluate the prognostic accuracy of a deep learning algorithm (SOFIA [Systematic Objective Fibrotic Imaging Analysis Algorithm]), trained and validated in the identification of usual interstitial pneumonia (UIP)-like features on HRCT (UIP probability), in a large cohort of well-characterized patients with progressive fibrotic lung disease drawn from a national registry. Methods: SOFIA and radiologist UIP probabilities were converted to Prospective Investigation of Pulmonary Embolism Diagnosis (PIOPED)-based UIP probability categories (UIP not included in the differential, 0-4%; low probability of UIP, 5-29%; intermediate probability of UIP, 30-69%; high probability of UIP, 70-94%; and pathognomonic for UIP, 95-100%), and their prognostic utility was assessed using Cox proportional hazards modeling. Measurements and Main Results: In multivariable analysis adjusting for age, sex, guideline-based radiologic diagnosis, anddisease severity (using total interstitial lung disease [ILD] extent on HRCT, percent predicted FVC, DlCO, or the composite physiologic index), only SOFIA UIP probability PIOPED categories predicted survival. SOFIA-PIOPED UIP probability categories remained prognostically significant in patients considered indeterminate (n = 83) by expert radiologist consensus (hazard ratio, 1.73; P < 0.0001; 95% confidence interval, 1.40-2.14). In patients undergoing surgical lung biopsy (n = 86), after adjusting for guideline-based histologic pattern and total ILD extent on HRCT, only SOFIA-PIOPED probabilities were predictive of mortality (hazard ratio, 1.75; P < 0.0001; 95% confidence interval, 1.37-2.25). Conclusions: Deep learning-based UIP probability on HRCT provides enhanced outcome prediction in patients with progressive fibrotic lung disease when compared with expert radiologist evaluation or guideline-based histologic pattern. In principle, this tool may be useful in multidisciplinary characterization of fibrotic lung disease. The utility of this technology as a decision support system when ILD expertise is unavailable requires further investigation.

Integration and Application of Clinical Practice Guidelines for the Diagnosis of Idiopathic Pulmonary Fibrosis and Fibrotic Hypersensitivity Pneumonitis.

Recent clinical practice guidelines have addressed the diagnosis of idiopathic pulmonary fibrosis (IPF) and fibrotic hypersensitivity pneumonitis (fHP). These disease-specific guidelines were developed independently, without clear direction on how to apply their respective recommendations concurrently within a single patient, where discrimination between these two fibrotic interstitial lung diseases represents a frequent diagnostic challenge. The objective of this review, created by an international group of experts, was to suggest a pragmatic approach on how to apply existing guidelines to distinguish IPF and fHP. Key clinical, radiologic, and pathologic features described in previous guidelines are integrated in a set of diagnostic algorithms, which then are placed in the broader context of multidisciplinary discussion to guide the generation of a consensus diagnosis. Although these algorithms necessarily reflect some uncertainty wherever strong evidence is lacking, they provide insight into the current approach favored by experts in the field based on currently available knowledge. The authors further identify priorities for future research to clarify ongoing uncertainties in the diagnosis of fibrotic interstitial lung diseases.

Integrating Clinical Probability into the Diagnostic Approach to Idiopathic Pulmonary Fibrosis: An International Working Group Perspective.

Background: When considering the diagnosis of idiopathic pulmonary fibrosis (IPF), experienced clinicians integrate clinical features that help to differentiate IPF from other fibrosing interstitial lung diseases, thus generating a "pre-test" probability of IPF. The aim of this international working group perspective was to summarize these features using a tabulated approach similar to chest HRCT and histopathologic patterns reported in the international guidelines for the diagnosis of IPF, and to help formally incorporate these clinical likelihoods into diagnostic reasoning to facilitate the diagnosis of IPF. Methods: The committee group identified factors that influence the clinical likelihood of a diagnosis of IPF, which was categorized as a pre-test clinical probability of IPF into "high" (70-100%), "intermediate" (30-70%), or "low" (0-30%). After integration of radiological and histopathological features, the post-test probability of diagnosis was categorized into "definite" (90-100%), "high confidence" (70-89%), "low confidence" (51-69%), or "low" (0-50%) probability of IPF. Findings: A conceptual Bayesian framework was created, integrating the clinical likelihood of IPF ("pre-test probability of IPF") with the HRCT pattern, the histopathology pattern when available, and/or the pattern of observed disease behavior, into a "post-test probability of IPF." The diagnostic probability of IPF was expressed using an adapted diagnostic ontology for fibrotic interstitial lung diseases. Interpretation: The present approach will help incorporate the clinical judgment into the diagnosis of IPF, thus facilitating the application of IPF diagnostic guidelines and, ultimately improving diagnostic confidence and reducing the need for invasive diagnostic techniques.

Essential Features of an Interstitial Lung Disease Multidisciplinary Meeting: An International Delphi Survey.

Rationale: The interstitial lung disease (ILD) multidisciplinary meetings (MDM), composed of pulmonologists, radiologists, and pathologists, is integral to the rendering of an accurate ILD diagnosis. However, there is significant heterogeneity in the conduct of ILD MDMs, and questions regarding their best practices remain unanswered. Objectives: To achieve consensus among ILD experts on essential components of an ILD MDM. Methods: Using a Delphi methodology, semi-structured interviews with ILD experts were used to identify key themes and features of ILD MDMs. These items informed two subsequent rounds of online questionnaires that were used to achieve consensus among a broader, international panel of ILD experts. Experts were asked to rate their level of agreement on a five-point Likert scale. An a priori threshold for consensus was set at a median score 4 or 5 with an interquartile range of 0. Results: We interviewed 15 ILD experts, and 102 ILD experts participated in the online questionnaires. Five items and two exploratory statements achieved consensus on being essential for an ILD MDM following two questionnaire rounds. There was consensus that the presence of at least one radiologist, a quiet setting with a visual projection system, a high-quality chest high-resolution computed tomography, and a standardized template summarizing collated patient data are essential components of an ILD MDM. Experts also agreed that it would be useful for ILD MDMs to undergo an annual benchmarking process and a validation process by fulfilling a minimum number of cases annually. Twenty-seven additional features were considered to be either highly desirable or desirable features based on the degree of consensus. Although our findings on desirable features are similar to the current literature, several of these remain controversial and warrant further research. The study also showed an agreement among participants on several future concepts to improve the ILD MDM, such as performing regular self-assessments and conducting research into shared practices to develop an international expert guideline statement on ILD MDMs. Conclusions: This Delphi study showed consensus among international ILD experts on essential and desirable features of an ILD MDM. Our data represent an important step toward potential collaborative research into future standardization of ILD MDMs.