Deep learning in interstitial lung disease-how long until daily practice

U-net convolutional neural network applied to progressive fibrotic interstitial lung disease: Is progression at CT scan associated with a clinical outcome?

BACKGROUND

Computational advances in artificial intelligence have led to the recent emergence of U-Net convolutional neural networks (CNNs) applied to medical imaging. Our objectives were to assess the progression of fibrotic interstitial lung disease (ILD) using routine CT scans processed by a U-Net CNN developed by our research team, and to identify a progression threshold indicative of poor prognosis.

METHODS

CT scans and clinical history of 32 patients with idiopathic fibrotic ILDs were retrospectively reviewed. Successive CT scans were processed by the U-Net CNN and ILD quantification was obtained. Correlation between ILD and FVC changes was assessed. ROC curve was used to define a threshold of ILD progression rate (PR) to predict poor prognostic (mortality or lung transplantation). The PR threshold was used to compare the cohort survival with Kaplan Mayer curves and log-rank test.

RESULTS

The follow-up was 3.8 ± 1.5 years encompassing 105 CT scans, with 3.3 ± 1.1 CT scans per patient. A significant correlation between ILD and FVC changes was obtained (p = 0.004, ρ = -0.30 [95% CI: -0.16 to -0.45]). Sixteen patients (50%) experienced unfavorable outcome including 13 deaths and 3 lung transplantations. ROC curve analysis showed an aera under curve of 0.83 (p < 0.001), with an optimal cut-off PR value of 4%/year. Patients exhibiting a PR ≥ 4%/year during the first two years had a poorer prognosis (p = 0.001).

CONCLUSIONS

Applying a U-Net CNN to routine CT scan allowed identifying patients with a rapid progression and unfavorable outcome.

Pulmonary Diseases in Older Patients: Understanding and Addressing the Challenges

As the global population ages, pulmonary diseases among older people have emerged as a significant and growing public health concern. The increasing incidence of these conditions has led to higher rates of morbidity and mortality among older adults. This perspective study offers a thorough overview of the prevalent pulmonary diseases affecting the elderly demographic. It delves into the challenges encountered during the diagnosis and management of these conditions in older individuals, considering factors such as comorbidities, functional limitations, and medication complexities. Furthermore, innovative strategies and personalized interventions such as precision medicine, advanced therapies, telemedicine solutions, and patient-centered support systems aimed at enhancing the care provided to older individuals grappling with pulmonary disorders are thoroughly explored. By addressing the unique needs and complexities of this vulnerable population, healthcare systems can strive towards improving outcomes and enhancing the quality of life for elderly individuals affected by pulmonary diseases.

External validation of Fibresolve, a machine-learning algorithm, to non-invasively diagnose idiopathic pulmonary fibrosis

BACKGROUND

Previous work has shown the ability of Fibresolve, a machine learning system, to non-invasively classify idiopathic pulmonary fibrosis (IPF) with a pre-invasive sensitivity of 53% and specificity of 86% versus other types of interstitial lung disease. Further external validation for the use of Fibresolve to classify IPF in patients with non-definite usual interstitial pneumonia (UIP) is needed. The aim of this study is to assess the sensitivity for Fibresolve to positively classify IPF in an external cohort of patients with a non-definite UIP radiographic pattern.

METHODS

This is a retrospective analysis of patients (n = 193) enrolled in two prospective phase two clinical trials that enrolled patients with IPF. We retrospectively identified patients with non-definite UIP on HRCT (n = 51), 47 of whom required surgical lung biopsy for diagnosis. Fibresolve was used to analyze the HRCT chest imaging which was obtained prior to invasive biopsy and sensitivity for final diagnosis of IPF was calculated.

RESULTS

The sensitivity of Fibresolve for the non-invasive classification of IPF in patients with a non-definite UIP radiographic pattern by HRCT was 76.5% (95% CI 66.5-83.7). For the subgroup of 47 patients who required surgical biopsy to aid in final diagnosis of IPF, Fibresolve had a sensitivity of 74.5% (95% CI 60.5-84.7).

CONCLUSION

In patients with suspected IPF with non-definite UIP on HRCT, Fibresolve can positively identify cases of IPF with high sensitivity. These results suggest that in combination with standard clinical assessment, Fibresolve has the potential to serve as an adjunct in the non-invasive diagnosis of IPF.

A Survey on Artificial Intelligence in Pulmonary Imaging

Over the last decade, deep learning (DL) has contributed a paradigm shift in computer vision and image recognition creating widespread opportunities of using artificial intelligence in research as well as industrial applications. DL has been extensively studied in medical imaging applications, including those related to pulmonary diseases. Chronic obstructive pulmonary disease, asthma, lung cancer, pneumonia, and, more recently, COVID-19 are common lung diseases affecting nearly 7.4% of world population. Pulmonary imaging has been widely investigated toward improving our understanding of disease etiologies and early diagnosis and assessment of disease progression and clinical outcomes. DL has been broadly applied to solve various pulmonary image processing challenges including classification, recognition, registration, and segmentation. This paper presents a survey of pulmonary diseases, roles of imaging in translational and clinical pulmonary research, and applications of different DL architectures and methods in pulmonary imaging with emphasis on DL-based segmentation of major pulmonary anatomies such as lung volumes, lung lobes, pulmonary vessels, and airways as well as thoracic musculoskeletal anatomies related to pulmonary diseases.

Treatable Traits in Systemic Sclerosis

Systemic sclerosis (SSc) is a chronic systemic disease within the spectrum of connective tissue diseases, specifically characterized by vascular abnormalities and inflammatory and fibrotic involvement of the skin and internal organs resulting in high morbidity and mortality. The clinical phenotype of SSc is heterogeneous, and serum autoantibodies together with the extent of skin involvement have a predictive value in the risk stratification. Current recommendations include an organ-based management according to the predominant involvement with only limited individual factors included in the treatment algorithm. Similar to what has been proposed for other chronic diseases, we hypothesize that a "treatable trait" approach based on relevant phenotypes and endotypes could address the unmet needs in SSc stratification and treatment to maximize the outcomes. We provide herein a comprehensive review and a critical discussion of the literature regarding potential treatable traits in SSc, focusing on established and candidate biomarkers, with the purpose of setting the bases for a precision medicine-based approach. The discussion, structured based on the organ involvement, allows to conjugate the pathogenetic mechanisms of tissue injury with the proposed predictors, particularly autoantibodies and other serum biomarkers. Ultimately, we are convinced that precision medicine is the ideal guide to manage a complex condition such as SSc for which available treatments are largely unsatisfactory.

Semi-Supervised Segmentation of Interstitial Lung Disease Patterns from CT Images via Self-Training with Selective Re-Training

Accurate segmentation of interstitial lung disease (ILD) patterns from computed tomography (CT) images is an essential prerequisite to treatment and follow-up. However, it is highly time-consuming for radiologists to pixel-by-pixel segment ILD patterns from CT scans with hundreds of slices. Consequently, it is hard to obtain large amounts of well-annotated data, which poses a huge challenge for data-driven deep learning-based methods. To alleviate this problem, we propose an end-to-end semi-supervised learning framework for the segmentation of ILD patterns (ESSegILD) from CT images via self-training with selective re-training. The proposed ESSegILD model is trained using a large CT dataset with slice-wise sparse annotations, i.e., only labeling a few slices in each CT volume with ILD patterns. Specifically, we adopt a popular semi-supervised framework, i.e., Mean-Teacher, that consists of a teacher model and a student model and uses consistency regularization to encourage consistent outputs from the two models under different perturbations. Furthermore, we propose introducing the latest self-training technique with a selective re-training strategy to select reliable pseudo-labels generated by the teacher model, which are used to expand training samples to promote the student model during iterative training. By leveraging consistency regularization and self-training with selective re-training, our proposed ESSegILD can effectively utilize unlabeled data from a partially annotated dataset to progressively improve the segmentation performance. Experiments are conducted on a dataset of 67 pneumonia patients with incomplete annotations containing over 11,000 CT images with eight different lung patterns of ILDs, with the results indicating that our proposed method is superior to the state-of-the-art methods.

A review on lung disease recognition by acoustic signal analysis with deep learning networks

Recently, assistive explanations for difficulties in the health check area have been made viable thanks in considerable portion to technologies like deep learning and machine learning. Using auditory analysis and medical imaging, they also increase the predictive accuracy for prompt and early disease detection. Medical professionals are thankful for such technological support since it helps them manage further patients because of the shortage of skilled human resources. In addition to serious illnesses like lung cancer and respiratory diseases, the plurality of breathing difficulties is gradually rising and endangering society. Because early prediction and immediate treatment are crucial for respiratory disorders, chest X-rays and respiratory sound audio are proving to be quite helpful together. Compared to related review studies on lung disease classification/detection using deep learning algorithms, only two review studies based on signal analysis for lung disease diagnosis have been conducted in 2011 and 2018. This work provides a review of lung disease recognition with acoustic signal analysis with deep learning networks. We anticipate that physicians and researchers working with sound-signal-based machine learning will find this material beneficial.

Interstitial lung disease diagnosis and prognosis using an AI system integrating longitudinal data

For accurate diagnosis of interstitial lung disease (ILD), a consensus of radiologic, pathological, and clinical findings is vital. Management of ILD also requires thorough follow-up with computed tomography (CT) studies and lung function tests to assess disease progression, severity, and response to treatment. However, accurate classification of ILD subtypes can be challenging, especially for those not accustomed to reading chest CTs regularly. Dynamic models to predict patient survival rates based on longitudinal data are challenging to create due to disease complexity, variation, and irregular visit intervals. Here, we utilize RadImageNet pretrained models to diagnose five types of ILD with multimodal data and a transformer model to determine a patient's 3-year survival rate. When clinical history and associated CT scans are available, the proposed deep learning system can help clinicians diagnose and classify ILD patients and, importantly, dynamically predict disease progression and prognosis.

Deep Transfer Learning Techniques-Based Automated Classification and Detection of Pulmonary Fibrosis from Chest CT Images

Pulmonary Fibrosis (PF) is a non-curable chronic lung disease. Therefore, a quick and accurate PF diagnosis is imperative. In the present study, we aim to compare the performance of the six state-of-the-art Deep Transfer Learning techniques to classify patients accurately and perform abnormality localization in Computer Tomography (CT) scan images. A total of 2299 samples comprising normal and PF-positive CT images were preprocessed. The preprocessed images were split into training (75%), validation (15%), and test data (10%). These transfer learning models were trained and validated by optimizing the hyperparameters, such as the learning rate and the number of epochs. The optimized architectures have been evaluated with different performance metrics to demonstrate the consistency of the optimized model. At epoch 26, using an optimized learning rate of 0.0000625, the ResNet50v2 model achieved the highest training and validation accuracy (training = 99.92%, validation = 99.22%) and minimum loss (training = 0.00428, validation = 0.00683) for CT images. The experimental evaluation on the independent testing data confirms that optimized ResNet50v2 outperformed every other optimized architecture under consideration achieving a perfect score of 1.0 in each of the standard performance measures such as accuracy, precision, recall, F1-score, Mathew Correlation Coefficient (MCC), Area under the Receiver Operating Characteristic (ROC-AUC) curve, and the Area under the Precision recall (AUC_PR) curve. Therefore, we can propose that the optimized ResNet50v2 is a reliable diagnostic model for automatically classifying PF-positive patients using chest CT images.

Post-COVID-19 interstitial lung disease: Insights from a machine learning radiographic model

Introduction

Post-acute sequelae of COVID-19 seem to be an emerging global crisis. Machine learning radiographic models have great potential for meticulous evaluation of post-COVID-19 interstitial lung disease (ILD).

Methods

In this multicenter, retrospective study, we included consecutive patients that had been evaluated 3 months following severe acute respiratory syndrome coronavirus 2 infection between 01/02/2021 and 12/5/2022. High-resolution computed tomography was evaluated through Imbio Lung Texture Analysis 2.1.

Results

Two hundred thirty-two (n = 232) patients were analyzed. FVC% predicted was ≥80, between 60 and 79 and <60 in 74.2% (n = 172), 21.1% (n = 49), and 4.7% (n = 11) of the cohort, respectively. DLCO% predicted was ≥80, between 60 and 79 and <60 in 69.4% (n = 161), 15.5% (n = 36), and 15.1% (n = 35), respectively. Extent of ground glass opacities was ≥30% in 4.3% of patients (n = 10), between 5 and 29% in 48.7% of patients (n = 113) and <5% in 47.0% of patients (n = 109). The extent of reticulation was ≥30%, 5-29% and <5% in 1.3% (n = 3), 24.1% (n = 56), and 74.6% (n = 173) of the cohort, respectively. Patients (n = 13, 5.6%) with fibrotic lung disease and persistent functional impairment at the 6-month follow-up received antifibrotics and presented with an absolute change of +10.3 (p = 0.01) and +14.6 (p = 0.01) in FVC% predicted at 3 and 6 months after the initiation of antifibrotic.

Conclusion

Post-COVID-19-ILD represents an emerging entity. A substantial minority of patients presents with fibrotic lung disease and might experience benefit from antifibrotic initiation at the time point that fibrotic-like changes are "immature." Machine learning radiographic models could be of major significance for accurate radiographic evaluation and subsequently for the guidance of therapeutic approaches.

Quantitative Assessment and Comparative Analysis of Longitudinal Lung CT Scans of Chest-Irradiated Nonhuman Primates

Computed tomography (CT) imaging has been used to diagnose radiation-induced lung injury for decades. However, histogram-based quantitative tools have rarely been applied to assess lung abnormality due to radiation-induced lung injury (RILI). Here, we used first-order summary statistics to derive and assess threshold measures extracted from whole lung histograms of CT radiodensity in rhesus macaques. For the present study, CT scans of animals exposed to 10 Gy of whole thorax irradiation were utilized from a previous study spanning 2-9 months postirradiation. These animals were grouped into survivors and non-survivors based on their clinical and experimental endpoints. We quantified the change in lung attenuation after irradiation relative to baseline using three density parameters; average lung density (ALD), percent change in hyper-dense lung volume (PCHV), hyperdense volume as a percent of total volume (PCHV/TV) at 2-month intervals and compared each parameter between the two irradiated groups (non-survivors and survivors). We also correlated our results with histological findings. All the three indices (ALD, PCHV, PCHV/TV) obtained from density histograms showed a significant increase in lung injury in non-survivors relative to survivors, with PCHV relatively more sensitive to detect early RILI changes. We observed a significant positive correlation between histologic pneumonitis scores and each of the three CT measurements, indicating that CT density is useful as a surrogate for histologic disease severity in RILI. CT-based three density parameters, ALD, PCHV, PCHV/TV, may serve as surrogates for likely histopathology patterns in future studies of RILI disease progression.

Artificial intelligence-based diagnosis of asbestosis: analysis of a database with applicants for asbestosis state aid

OBJECTIVES

In many countries, workers who developed asbestosis due to their occupation are eligible for government support. Based on the results of clinical examination, a team of pulmonologists determine the eligibility of patients to these programs. In this Dutch cohort study, we aim to demonstrate the potential role of an artificial intelligence (AI)-based system for automated, standardized, and cost-effective evaluation of applications for asbestosis patients.

METHODS

A dataset of n = 523 suspected asbestosis cases/applications from across the Netherlands was retrospectively collected. Each case/application was reviewed, and based on the criteria, a panel of three pulmonologists would determine eligibility for government support. An AI system is proposed, which uses thoracic CT images as input, and predicts the assessment of the clinical panel. Alongside imaging, we evaluated the added value of lung function parameters.

RESULTS

The proposed AI algorithm reached an AUC of 0.87 (p < 0.001) in the prediction of accepted versus rejected applications. Diffusion capacity (DLCO) also showed comparable predictive value (AUC = 0.85, p < 0.001), with little correlation between the two parameters (r-squared = 0.22, p < 0.001). The combination of the imaging AI score and DLCO achieved superior performance (AUC = 0.95, p < 0.001). Interobserver variability between pulmonologists on the panel was estimated at alpha = 0.65 (Krippendorff's alpha).

CONCLUSION

We developed an AI system to support the clinical decision-making process for the application to the government support for asbestosis. A multicenter prospective validation study is currently ongoing to examine the added value and reliability of this system alongside the clinic panel.

KEY POINTS

• Artificial intelligence can detect imaging patterns of asbestosis in CT scans in a cohort of patients applying for state aid. • Combining the AI prediction with the diffusing lung function parameter reaches the highest diagnostic performance. • Specific cases with fibrosis but no asbestosis were correctly classified, suggesting robustness of the AI system, which is currently under prospective validation.

Enhancing Imagistic Interstitial Lung Disease Diagnosis by Using Complex Networks

Background and Objectives: Diffuse interstitial lung diseases (DILD) are a heterogeneous group of over 200 entities, some with dramatical evolution and poor prognostic. Because of their overlapping clinical, physiopathological and imagistic nature, successful management requires early detection and proper progression evaluation. This paper tests a complex networks (CN) algorithm for imagistic aided diagnosis fitness for the possibility of achieving relevant and novel DILD management data. Materials and Methods: 65 DILD and 31 normal high resolution computer tomography (HRCT) scans were selected and analyzed with the CN model. Results: The algorithm is showcased in two case reports and then statistical analysis on the entire lot shows that a CN algorithm quantifies progression evaluation with a very fine accuracy, surpassing functional parameters’ variations. The CN algorithm can also be successfully used for early detection, mainly on the ground glass opacity Hounsfield Units band of the scan. Conclusions: A CN based computer aided diagnosis could provide the much-required data needed to successfully manage DILDs.

A Novel Method for Lung Image Processing Using Complex Networks

The High-Resolution Computed Tomography (HRCT) detection and diagnosis of diffuse lung disease is primarily based on the recognition of a limited number of specific abnormal findings, pattern combinations or their distributions, as well as anamnesis and clinical information. Since texture recognition has a very high accuracy percentage if a complex network approach is used, this paper aims to implement such a technique customized for diffuse interstitial lung diseases (DILD). The proposed procedure translates HRCT lung imaging into complex networks by taking samples containing a secondary lobule, converting them into complex networks and analyzing them in three dimensions: emphysema, ground glass opacity, and consolidation. This method was evaluated on a 60-patient lot and the results showed a clear, quantifiable difference between healthy and affected lungs. By deconstructing the image on three pathological axes, the method offers an objective way to quantify DILD details which, so far, have only been analyzed subjectively.

Quantitative CT and machine learning classification of fibrotic interstitial lung diseases

OBJECTIVES

To evaluate quantitative computed tomography (QCT) features and QCT feature-based machine learning (ML) models in classifying interstitial lung diseases (ILDs). To compare QCT-ML and deep learning (DL) models' performance.

METHODS

We retrospectively identified 1085 patients with pathologically proven usual interstitial pneumonitis (UIP), nonspecific interstitial pneumonitis (NSIP), and chronic hypersensitivity pneumonitis (CHP) who underwent peri-biopsy chest CT. Kruskal-Wallis test evaluated QCT feature associations with each ILD. QCT features, patient demographics, and pulmonary function test (PFT) results trained eXtreme Gradient Boosting (training/validation set n = 911) yielding 3 models: M1 = QCT features only; M2 = M1 plus age and sex; M3 = M2 plus PFT results. A DL model was also developed. ML and DL model areas under the receiver operating characteristic curve (AUC) and 95% confidence intervals (CIs) were compared for multiclass (UIP vs. NSIP vs. CHP) and binary (UIP vs. non-UIP) classification performances.

RESULTS

The majority (69/78 [88%]) of QCT features successfully differentiated the 3 ILDs (adjusted p ≤ 0.05). All QCT-ML models achieved higher AUC than the DL model (multiclass AUC micro-averages 0.910, 0.910, 0.925, and 0.798 and macro-averages 0.895, 0.893, 0.925, and 0.779 for M1, M2, M3, and DL respectively; binary AUC 0.880, 0.899, 0.898, and 0.869 for M1, M2, M3, and DL respectively). M3 demonstrated statistically significant better performance compared to M2 (∆AUC: 0.015, CI: [0.002, 0.029]) for multiclass prediction.

CONCLUSIONS

QCT features successfully differentiated pathologically proven UIP, NSIP, and CHP. While QCT-based ML models outperformed a DL model for classifying ILDs, further investigations are warranted to determine if QCT-ML, DL, or a combination will be superior in ILD classification.

KEY POINTS

• Quantitative CT features successfully differentiated pathologically proven UIP, NSIP, and CHP. • Our quantitative CT-based machine learning models demonstrated high performance in classifying UIP, NSIP, and CHP histopathology, outperforming a deep learning model. • While our quantitative CT-based machine learning models performed better than a DL model, additional investigations are needed to determine whether either or a combination of both approaches delivers superior diagnostic performance.

Multidisciplinary clinical guidance on trastuzumab deruxtecan (T-DXd)-related interstitial lung disease/pneumonitis-Focus on proactive monitoring, diagnosis, and management

Trastuzumab deruxtecan (T-DXd; DS-8201) is an antibody-drug conjugate targeting human epidermal growth factor receptor 2. Interstitial lung disease (ILD)/pneumonitis is an adverse event associated with T-DXd; in most cases, it is low grade (grade ≤ 2) and can be treated effectively but may develop to be fatal in some instances. It is important to increase patient and provider understanding of T-DXd-related ILD/pneumonitis to improve patient outcomes. Drug-related ILD/pneumonitis is a diagnosis of exclusion; other possible causes of lung injury/imaging findings must be ruled out for an accurate diagnosis. Symptoms can be nonspecific, and identifying early symptoms is challenging; therefore, diagnosis is often delayed. We reviewed characteristics of patients who developed T-DXd-related ILD/pneumonitis and its patterns, produced multidisciplinary guidelines on diagnosis and management, and described areas for future investigation. Ongoing studies are collecting data on T-DXd-related ILD/pneumonitis to further our understanding of its clinical patterns and mechanisms. SEARCH STRATEGY AND SELECTION CRITERIA: References were identified based on the guidelines used by the authors in treating interstitial lung disease and pneumonitis. Searches of the authors' own files were also completed. A search of PubMed with the search terms (trastuzumab deruxtecan) AND (interstitial lung disease) AND (guidelines) was conducted on November 1, 2021, with no restrictions based on publication date, and the two articles yielded by the search were included.

The role of precision medicine in interstitial lung disease

The management of interstitial lung disease (ILD) may benefit from a conceptual shift. Increased understanding of this complex and heterogeneous group of disorders over the past 20 years has highlighted the need for individualised treatment strategies that encompass diagnostic classification and disease behaviour. Biomarker-based approaches to precision medicine hold the greatest promise. Robust, large-scale biomarker-based technologies supporting ILD diagnosis have been developed, and future applications relating to staging, prognosis and assessment of treatment response are emerging. Artificial intelligence may redefine our ability to base prognostic evaluation on both diagnosis and underlying disease processes, sharpening individualised treatment algorithms to a level not previously achieved. Compared with therapeutic areas such as oncology, precision medicine in ILD is still in its infancy. However, the heterogeneous nature of ILD suggests that many relevant molecular, environmental and behavioural targets may serve as useful biomarkers if we are willing to invest in their identification and validation.

Evaluation of a Novel Content-Based Image Retrieval System for the Differentiation of Interstitial Lung Diseases in CT Examinations

To evaluate the reader's diagnostic performance against the ground truth with and without the help of a novel content-based image retrieval system (CBIR) that retrieves images with similar CT patterns from a database of 79 different interstitial lung diseases. We evaluated three novice readers' and three resident physicians' (with at least three years of experience) diagnostic performance evaluating 50 different CTs featuring 10 different patterns (e.g., honeycombing, tree-in bud, ground glass, bronchiectasis, etc.) and 24 different diseases (sarcoidosis, UIP, NSIP, Aspergillosis, COVID-19 pneumonia etc.). The participants read the cases first without assistance (and without feedback regarding correctness), and with a 2-month interval in a random order with the assistance of the novel CBIR. To invoke the CBIR, a ROI is placed into the pathologic pattern by the reader and the system retrieves diseases with similar patterns. To further narrow the differential diagnosis, the readers can consult an integrated textbook and have the possibility of selecting high-level semantic features representing clinical information (chronic, infectious, smoking status, etc.). We analyzed readers' accuracy without and with CBIR assistance and further tested the hypothesis that the CBIR would help to improve diagnostic performance utilizing Wilcoxon signed rank test. The novice readers demonstrated an unassisted accuracy of 18/28/44%, and an assisted accuracy of 84/82/90%, respectively. The resident physicians demonstrated an unassisted accuracy of 56/56/70%, and an assisted accuracy of 94/90/96%, respectively. For each reader, as well as overall, Sign test demonstrated statistically significant (p < 0.01) difference between the unassisted and the assisted reads. For students and physicians, Chi²-test and Mann-Whitney-U test demonstrated statistically significant (p < 0.01) difference for unassisted reads and statistically insignificant (p > 0.01) difference for assisted reads. The evaluated CBIR relying on pattern analysis and featuring the option to filter the results of the CBIR by predominant characteristics of the diseases via selecting high-level semantic features helped to drastically improve novices' and resident physicians' accuracy in diagnosing interstitial lung diseases in CT.

Safety and Effectiveness of Mycophenolate Mofetil in Interstitial Lung Diseases: Insights from a Machine Learning Radiographic Model

INTRODUCTION

Treatment of interstitial lung diseases (ILDs) other than idiopathic pulmonary fibrosis (IPF) often includes systemic corticosteroids. Use of steroid-sparing agents is amenable to avoid potential side effects.

METHODS

Functional indices and high-resolution computed tomography (HRCT) patterns of patients with non-IPF ILDs receiving mycophenolate mofetil (MMF) with a minimum follow-up of 1 year were analyzed. Two independent radiologists and a machine learning software system (Imbio 1.4.2.) evaluated HRCT patterns.

RESULTS

Fifty-five (n = 55) patients were included in the analysis (male: 30 [55%], median age: 65.0 [95% CI: 59.7-70.0], mean forced vital capacity %predicted [FVC %pred.] ± standard deviation [SD]: 69.4 ± 18.3, mean diffusing capacity of lung for carbon monoxide %pred. ± SD: 40.8 ± 14.3, hypersensitivity pneumonitis: 26, connective tissue disease-ILDs [CTD-ILDs]: 22, other ILDs: 7). There was no significant difference in mean FVC %pred. post-6 months (1.59 ± 2.04) and 1 year (-0.39 ± 2.49) of treatment compared to baseline. Radiographic evaluation showed no significant difference between baseline and post-1 year %ground glass opacities (20.0 [95% CI: 14.4-30.0] vs. 20.0 [95% CI: 14.4-25.6]) and %reticulation (5.0 [95% CI: 2.0-15.6] vs. 7.5 [95% CI: 2.0-17.5]). A similar performance between expert radiologists and Imbio software analysis was observed in assessing ground glass opacities (intraclass correlation coefficient [ICC] = 0.73) and reticulation (ICC = 0.88). Fourteen patients (25.5%) reported at least one side effect and 8 patients (14.5%) switched to antifibrotics due to disease progression.

CONCLUSION

Our data suggest that MMF is a safe and effective steroid-sparing agent leading to disease stabilization in a proportion of patients with non-IPF ILDs. Machine learning software systems may exhibit similar performance to specialist radiologists and represent fruitful diagnostic and prognostic tools.

[Radiological features of interstitial lung diseases]

In addition to pneumology and pathology, radiology is an essential discipline in the interdisciplinary diagnosis of interstitial lung diseases (ILDs). The gold standard for diagnosis of ILD is computed tomography. Diagnostic findings are based on specific radiological signs such as interlobular septal thickening and nodular changes. From these signs and their distribution within the lung, radiological patterns can be derived, e.g., usual interstitial pneumonia, nonspecific interstitial pneumonia, or organizing pneumonia. Various differential diagnoses result from the radiological pattern, which can then be further limited in an interdisciplinary manner with the clinic and pathology and, if necessary, trigger further diagnostics.The visual assessment of interstitial lung changes requires experience and training and is nevertheless error-prone with high inter- and intraobserver variabilities. Recently, therefore, computer-aided analysis of ILDs has been increasingly promoted. These computer programs analyze the density distribution of the lung parenchyma using parameters such as mean lung density, skewness, and kurtosis thus enabling the quantification and assessment of the course of disease. Furthermore, texture analysis and artificial intelligence are used to characterize parenchymal changes and differentiate between regions of ground glass, reticulation, and honeycombing. Modern dual-energy CT methods allow a combined, regional recording of both the morphology and the function and provide information about regional ventilation and perfusion.