AI Application in Biomedical Signal and Image Analysis
Margaliot, Shai |
Recenzentas / Reviewer |
Šis darbas nagrinėja dirbtinio intelekto taikymą širdies garsų įrašų analizėje naudojant fonokardiogramas, siekiant aptikti ūžesius ir pagerinti diagnostiką. Jame apžvelgiamas ir analizuojamas „PhysioNet Challenge 2022“ konkursas, vertinami komandų sukurti algoritmai, ypatingą dėmesį skiriant LSMU komandai, ir siūlomi patobulinimai, pagrįsti bangaine signalo apdorojimu ir CNN (konvoliuciniais neuroniniais tinklais). Rezultatai patvirtina neinvazinių dirbtiniu intelektu pagrįstų diagnostikos priemonių kūrimo galimybes.
In recent years, the improvement of new methodologies for analysing enormous amounts of high complexity biomedical data has led to an increasing number of algorithms based on Artificial Intelligence (AI) technology, including Machine Learning and Artificial Neural Networks. Applications of Al in medical practice are capable of analysing diverse medical data and detecting numerous patterns of diagnostically important information in less time and at higher precision when compared to human expert. Open access data bases have unrestricted access health data which is well-characterized and usually come from wide range of studies. Their mission is to conduct and catalyse biomedical research and education. One of the open access biomedical data base is PhysioNet (www.physionet.org). PhysioNet in cooperation with the annual conference "Computing in Cardiology" (www.cinc.org) hosts an annual series of Challenges, focusing research on unsolved problems in clinical and basic science. In the year 2022 Challenge was - "Heart Murmur Detection from Phonocardiogram Recordings: The George B. Moody PhysioNet Challenge 2022". The Challenge revealed a lot of ideas for further development of algorithms for solving very important and actual clinical problem. Detail analysis of published methods and their results could reveal most perspective ways of future development of methods and algorithms for Phonocardiogram signal analysis in regard to locally available resources.