Deep Learning-Based Object Detection Strategies for Disease Detection and Localization in Chest X-Ray Images

Background and Objectives: Chest X-ray (CXR) images are commonly used to diagnose respiratory and cardiovascular diseases. However, traditional manual interpretation is often subjective, time-consuming, and prone to errors, leading to inconsistent detection accuracy and poor generalization. In this...

Full description

Saved in:
Bibliographic Details
Main Authors: Yi-Ching Cheng, Yi-Chieh Hung, Guan-Hua Huang, Tai-Been Chen, Nan-Han Lu, Kuo-Ying Liu, Kuo-Hsuan Lin
Format: Article
Language:English
Published: MDPI AG 2024-11-01
Series:Diagnostics
Subjects:
Online Access:https://www.mdpi.com/2075-4418/14/23/2636
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1850061254183354368
author Yi-Ching Cheng
Yi-Chieh Hung
Guan-Hua Huang
Tai-Been Chen
Nan-Han Lu
Kuo-Ying Liu
Kuo-Hsuan Lin
author_facet Yi-Ching Cheng
Yi-Chieh Hung
Guan-Hua Huang
Tai-Been Chen
Nan-Han Lu
Kuo-Ying Liu
Kuo-Hsuan Lin
author_sort Yi-Ching Cheng
collection DOAJ
description Background and Objectives: Chest X-ray (CXR) images are commonly used to diagnose respiratory and cardiovascular diseases. However, traditional manual interpretation is often subjective, time-consuming, and prone to errors, leading to inconsistent detection accuracy and poor generalization. In this paper, we present deep learning-based object detection methods for automatically identifying and annotating abnormal regions in CXR images. Methods: We developed and tested our models using disease-labeled CXR images and location-bounding boxes from E-Da Hospital. Given the prevalence of normal images over diseased ones in clinical settings, we created various training datasets and approaches to assess how different proportions of background images impact model performance. To address the issue of limited examples for certain diseases, we also investigated few-shot object detection techniques. We compared convolutional neural networks (CNNs) and Transformer-based models to determine the most effective architecture for medical image analysis. Results: The findings show that background image proportions greatly influenced model inference. Moreover, schemes incorporating binary classification consistently improved performance, and CNN-based models outperformed Transformer-based models across all scenarios. Conclusions: We have developed a more efficient and reliable system for the automated detection of disease labels and location bounding boxes in CXR images.
format Article
id doaj-art-201bc6a7384b4bb29e3d67ed28f60784
institution DOAJ
issn 2075-4418
language English
publishDate 2024-11-01
publisher MDPI AG
record_format Article
series Diagnostics
spelling doaj-art-201bc6a7384b4bb29e3d67ed28f607842025-08-20T02:50:18ZengMDPI AGDiagnostics2075-44182024-11-011423263610.3390/diagnostics14232636Deep Learning-Based Object Detection Strategies for Disease Detection and Localization in Chest X-Ray ImagesYi-Ching Cheng0Yi-Chieh Hung1Guan-Hua Huang2Tai-Been Chen3Nan-Han Lu4Kuo-Ying Liu5Kuo-Hsuan Lin6Institute of Statistics, National Yang Ming Chiao Tung University, Hsinchu 300093, TaiwanInstitute of Statistics, National Yang Ming Chiao Tung University, Hsinchu 300093, TaiwanInstitute of Statistics, National Yang Ming Chiao Tung University, Hsinchu 300093, TaiwanDepartment of Radiological Technology, Faculty of Medical Technology, Teikyo University, Tokyo 173-8605, JapanDepartment of Radiology, E-Da Cancer Hospital, I-Shou University, Kaohsiung 824005, TaiwanDepartment of Radiology, E-Da Cancer Hospital, I-Shou University, Kaohsiung 824005, TaiwanDepartment of Emergency Medicine, E-Da Hospital, I-Shou University, Kaohsiung 824005, TaiwanBackground and Objectives: Chest X-ray (CXR) images are commonly used to diagnose respiratory and cardiovascular diseases. However, traditional manual interpretation is often subjective, time-consuming, and prone to errors, leading to inconsistent detection accuracy and poor generalization. In this paper, we present deep learning-based object detection methods for automatically identifying and annotating abnormal regions in CXR images. Methods: We developed and tested our models using disease-labeled CXR images and location-bounding boxes from E-Da Hospital. Given the prevalence of normal images over diseased ones in clinical settings, we created various training datasets and approaches to assess how different proportions of background images impact model performance. To address the issue of limited examples for certain diseases, we also investigated few-shot object detection techniques. We compared convolutional neural networks (CNNs) and Transformer-based models to determine the most effective architecture for medical image analysis. Results: The findings show that background image proportions greatly influenced model inference. Moreover, schemes incorporating binary classification consistently improved performance, and CNN-based models outperformed Transformer-based models across all scenarios. Conclusions: We have developed a more efficient and reliable system for the automated detection of disease labels and location bounding boxes in CXR images.https://www.mdpi.com/2075-4418/14/23/2636chest X-raysdeep learningfew-shot object detectionobject detection
spellingShingle Yi-Ching Cheng
Yi-Chieh Hung
Guan-Hua Huang
Tai-Been Chen
Nan-Han Lu
Kuo-Ying Liu
Kuo-Hsuan Lin
Deep Learning-Based Object Detection Strategies for Disease Detection and Localization in Chest X-Ray Images
Diagnostics
chest X-rays
deep learning
few-shot object detection
object detection
title Deep Learning-Based Object Detection Strategies for Disease Detection and Localization in Chest X-Ray Images
title_full Deep Learning-Based Object Detection Strategies for Disease Detection and Localization in Chest X-Ray Images
title_fullStr Deep Learning-Based Object Detection Strategies for Disease Detection and Localization in Chest X-Ray Images
title_full_unstemmed Deep Learning-Based Object Detection Strategies for Disease Detection and Localization in Chest X-Ray Images
title_short Deep Learning-Based Object Detection Strategies for Disease Detection and Localization in Chest X-Ray Images
title_sort deep learning based object detection strategies for disease detection and localization in chest x ray images
topic chest X-rays
deep learning
few-shot object detection
object detection
url https://www.mdpi.com/2075-4418/14/23/2636
work_keys_str_mv AT yichingcheng deeplearningbasedobjectdetectionstrategiesfordiseasedetectionandlocalizationinchestxrayimages
AT yichiehhung deeplearningbasedobjectdetectionstrategiesfordiseasedetectionandlocalizationinchestxrayimages
AT guanhuahuang deeplearningbasedobjectdetectionstrategiesfordiseasedetectionandlocalizationinchestxrayimages
AT taibeenchen deeplearningbasedobjectdetectionstrategiesfordiseasedetectionandlocalizationinchestxrayimages
AT nanhanlu deeplearningbasedobjectdetectionstrategiesfordiseasedetectionandlocalizationinchestxrayimages
AT kuoyingliu deeplearningbasedobjectdetectionstrategiesfordiseasedetectionandlocalizationinchestxrayimages
AT kuohsuanlin deeplearningbasedobjectdetectionstrategiesfordiseasedetectionandlocalizationinchestxrayimages