Automatic Respiratory Disease Detection and Classification from Lung Images: A Detailed Review

Shemy Syed, R Elakkiya*, N. E. Pears

*Corresponding author for this work

Research output: Contribution to journalArticlepeer-review


The image based automatic detection and classification of respiratory diseases has a large and complex range of solutions that span both computer vision and medical imaging literature. This has been driven by a range of powerful recent developments in deep learning, transformers, and new founding models, and these have significantly improved performances across a range of benchmarks. We provide a comprehensive review of these approaches, comparing and contrasting them in terms of performance, generalization, trading data requirements, and computational costs for both training and inference. Our review maps to the key processes found in the literature namely dataset and benchmark development and curation, data finetuning, segmentation, feature Extraction, localization, object detection masking, and classification of diseases. This will help locate the existing problems in each phase and efficiently find solutions. It will give an overview of recent advancements and potential future directions. This review article could be used as a quick guide by fellow researchers to get a detailed view of the problem domain and plan their contributions.
Original languageEnglish
Number of pages31
JournalJournal of Autonomous Intelligence
Publication statusAccepted/In press - 4 Dec 2023

Bibliographical note

This is an author-produced version of the published paper. Uploaded in accordance with the University’s Research Publications and Open Access policy.

Cite this