Explainable Deep Learning for Non-Invasive Detection of Pulmonary Artery Hypertension from Heart Sounds

• Type: Article in International Conference Proceedings Book
• Date: 2022

Authors

Gaudio, A (Author) Other; Coimbra, M (Author) FCUP; Aurélio Campilho (Author) FEUP; Smailagic, A (Author) Other; Schmidt, SE (Author) Other; Renna, F (Author) FCUP

Conference proceedings International

Title: Computing in Cardiology, CinC 2022, Tampere, Finland, September 4-7, 2022

Pages: 1-4

2022 Computing in Cardiology, CinC 2022

4 September 2022 through 7 September 2022

Indexing

Scopus

Other information

Authenticus ID: P-00Y-8W5

DOI: 10.22489/cinc.2022.295

Abstract (EN): Late diagnoses of patients affected by pulmonary artery hypertension (PH) have a poor outcome. This observation has led to a call for earlier, non-invasive PH detection. Cardiac auscultation offers a non-invasive and cost-effective alternative to both right heart catheterization and doppler analysis in analysis of PH. We propose to detect PH via analysis of digital heart sound recordings with over-parameterized deep neural networks. In contrast with previous approaches in the literature, we assess the impact of a pre-processing step aiming to separate S2 sound into the aortic (A2) and pulmonary (P2) components. We obtain an area under the ROC curve of. 95, improving over our adaptation of a state-of-the-art Gaussian mixture model PH detector by +.17. Post-hoc explanations and analysis show that the availability of separated A2 and P2 components contributes significantly to prediction. Analysis of stethoscope heart sound recordings with deep networks is an effective, low-cost and non-invasive solution for the detection of pulmonary hypertension. © 2022 Creative Commons.

Language: English

Type (Professor's evaluation): Scientific