Classification of pulmonary sounds through deep learning for the diagnosis of interstitial lung diseases secondary to connective tissue diseases.

Early diagnosis of interstitial lung diseases secondary to connective tissue diseases is critical for the treatment and survival of patients. The symptoms, like dry cough and dyspnea, appear late in the clinical history and are not specific, moreover, the current approach to confirm the diagnosis of interstitial lung disease is based on high resolution computer tomography. However, computer tomography involves x-ray exposure for patients and high costs for the Health System, therefore preventing its use for a massive screening campaign in elder people. In this work we investigate the use of deep learning techniques for the classification of pulmonary sounds acquired from patients affected by connective tissue diseases. The novelty of the work consists of a suitably developed pre-processing pipeline for de-noising and data augmentation. The proposed approach is combined with a clinical study where the ground truth is represented by high resolution computer tomography. Various convolutional neural networks have provided an overall accuracy as high as 91% in the classification of lung sounds and have led to an overwhelming diagnostic accuracy in the range 91%-93%. Modern high performance hardware for edge computing can easily support our algorithms. This solution paves the way for a vast screening campaign of interstitial lung diseases in elder people on the basis of a non-invasive and cheap thoracic auscultation.

VECTOR: An algorithm for the detection of COVID-19 pneumonia from velcro-like lung sounds.

The coronavirus disease 2019 (COVID-19) has severely stressed the sanitary systems of all countries in the world. One of the main issues that physicians are called to tackle is represented by the monitoring of pauci-symptomatic COVID-19 patients at home and, generally speaking, everyone the access to the hospital might or should be severely reduced. Indeed, the early detection of interstitial pneumonia is particularly relevant for the survival of these patients. Recent studies on rheumatoid arthritis and interstitial lung diseases have shown that pathological pulmonary sounds can be automatically detected by suitably developed algorithms. The scope of this preliminary work consists of proving that the pathological lung sounds evidenced in patients affected by COVID-19 pneumonia can be automatically detected as well by the same class of algorithms. In particular the software VECTOR, suitably devised for interstitial lung diseases, has been employed to process the lung sounds of 28 patient recorded in the emergency room at the university hospital of Modena (Italy) during December 2020. The performance of VECTOR has been compared with diagnostic techniques based on imaging, namely lung ultrasound, chest X-ray and high resolution computed tomography, which have been assumed as ground truth. The results have evidenced a surprising overall diagnostic accuracy of 75% even if the staff of the emergency room has not been suitably trained for lung auscultation and the parameters of the software have not been optimized to detect interstitial pneumonia. These results pave the way to a new approach for monitoring the pulmonary implication in pauci-symptomatic COVID-19 patients.

Usefulness of digital velcro crackles detection in identification of interstitial lung disease in patients with connective tissue diseases.

OBJECTIVES: This study aims to evaluate the diagnostic accuracy of the VECTOR software in patients with connective tissue diseases (CTDs), compared with the reference standard of high-resolution computed tomography (HRCT). PATIENTS AND METHODS: The study included 98 consecutive patients of CTD (24 males, 74 females; median age 66 years; range, 24 to 85 years) with a recent HRCT. Patients were evaluated in a blindly manner by VECTOR and the results obtained by the algorithm were compared with the presence of interstitial lung disease (ILD) according to HRCT. RESULTS: Interstitial lung disease was detected in 42.8% of subjects. VECTOR correctly classified 81/98 patients, with a diagnostic accuracy of 82.6%; sensitivity and specificity were 88.1% and 78.6%, respectively. Only 5/42 patients with ILD were not correctly classified by VECTOR, while false positive cases were 21.4%. No significant differences were observed according to the radiologic pattern of ILD. CONCLUSION: VECTOR showed high sensitivity, specificity and diagnostic accuracy, allowing selecting patients to be investigated with HRCT. The relatively high frequency rate of false positive results is acceptable if compared with the lack of effective screening methods for this complication of CTDs.

Erratum to acute exacerbation of interstitial lung diseases secondary to systemic rheumatic diseases: a prospective study and review of the literature.

[This corrects the article DOI: 10.21037/jtd.2019.03.28.].

"Velcro-type" crackles predict specific radiologic features of fibrotic interstitial lung disease.

BACKGROUND: "Velcro-type" crackles on chest auscultation are considered a typical acoustic finding of Fibrotic Interstitial Lung Disease (FILD), however whether they may have a role in the early detection of these disorders has been unknown. This study investigated how "Velcro-type" crackles correlate with the presence of distinct patterns of FILD and individual radiologic features of pulmonary fibrosis on High Resolution Computed Tomography (HRCT). METHODS: Lung sounds were digitally recorded from subjects immediately prior to undergoing clinically indicated chest HRCT. Audio files were independently assessed by two chest physicians and both full volume and single HRCT sections corresponding to the recording sites were extracted. The relationships between audible "Velcro-type" crackles and radiologic HRCT patterns and individual features of pulmonary fibrosis were investigated using multivariate regression models. RESULTS: 148 subjects were enrolled: bilateral "Velcro-type" crackles predicted the presence of FILD at HRCT (OR 13.46, 95% CI 5.85-30.96, p < 0.001) and most strongly the Usual Interstitial Pneumonia (UIP) pattern (OR 19.8, 95% CI 5.28-74.25, p < 0.001). Extent of isolated reticulation (OR 2.04, 95% CI 1.62-2.57, p < 0.001), honeycombing (OR 1.88, 95% CI 1.24-2.83, < 0.01), ground glass opacities (OR 1.74, 95% CI 1.29-2.32, p < 0.001) and traction bronchiectasis (OR 1.55, 95% CI 1.03-2.32, p < 0.05) were all independently associated with the presence of "Velcro-type" crackles. CONCLUSIONS: "Velcro-type" crackles predict the presence of FILD and directly correlate with the extent of distinct radiologic features of pulmonary fibrosis. Such evidence provides grounds for further investigation of lung sounds as an early identification tool in FILD.

Analysis of pulmonary sounds for the diagnosis of interstitial lung diseases secondary to rheumatoid arthritis.

The diagnosis of interstitial lung diseases in patients affected by rheumatoid arthritis is fundamental to improving their survival rate. In particular, the average survival time of patients affected by rheumatoid arthritis with pulmonary implications is approximately 3 years. The gold standard for confirming the diagnosis of this disease is computer tomography. However, it is very difficult to raise diagnosis suspicion because the symptoms of the disease are extremely common in elderly people. The detection of the so-called velcro crackle in lung sounds can effectively raise the suspicion of an interstitial disease and speed up diagnosis. However, this task largely relies on the experience of physicians and has not yet been standardized in clinical practice. The diagnosis of interstitial lung diseases based on thorax auscultation still represents an underexplored field in the study of rheumatoid arthritis. In this study, we investigate the problem of the automatic detection of velcro crackle in lung sounds. In practice, the patient is auscultated using a digital stethoscope and the lung sounds are saved to a file. The acquired digital data are then analysed using a suitably developed algorithm. In particular, the proposed solution relies on the empirical observation that the audio bandwidth associated with velcro crackle is larger than that associated with healthy breath sounds. Experimental results from a database of 70 patients affected by rheumatoid arthritis demonstrate that the developed tool can outperform specialized physicians in terms of diagnosing pulmonary disorders. The overall accuracy of the proposed solution is 90.0%, with negative and positive predictive values of 95.0% and 83.3%, respectively, whereas the reliability of physician diagnosis is in the range of 60-70%. The devised algorithm represents an enabling technology for a novel approach to the diagnosis of interstitial lung diseases in patients affected by rheumatoid arthritis.