No Detail Found!

Multi-stage Fine-tuning Approach for AI-based Chest X- ray Abnormality Detection

Author(s)
Abdulhadi Saeed Aljumayi Saudi Red Crescent Authority, Saudi Arabia
Abdullah Safar Al-Thaqafi Saudi Red Crescent Authority, Saudi Arabia
Abdulkarim Abed Alrabie Saudi Red Crescent Authority, Saudi Arabia
Omar Ibrahim Althobaiti Saudi Red Crescent Authority, Saudi Arabia
Rayed Mohammed Alotaibi Saudi Red Crescent Authority, Saudi Arabia
Ramzi Mohammed Al-Asimi Saudi Red Crescent Authority, Saudi Arabia
Ali Ahmad Ali Alzahrani Saudi Red Crescent Authority, Saudi Arabia
Mohammad Kalaf Ullah Salah Alharthi Saudi Red Crescent Authority, Saudi Arabia
MuflehMoishaNawarAlotibi Saudi Red Crescent Authority, Saudi Arabia
AbdulrahmanSwylhAlropeay Saudi Red Crescent Authority, Saudi Arabia
Oqab Muhayya Algethami Saudi Red Crescent Authority, Saudi Arabia
Abdullah Mohammed Al Nefaie Saudi Red Crescent Authority, Saudi Arabia
Abdulghafur Qar I Abdulrahman Saudi Red Crescent Authority, Saudi Arabia
Abstract
Correct detection and localization of thoracic diseases on chest Xray images supports or even secures an early diagnosis and treatment-oriented planning. This paper proposes a version of the YOLOv5 deep learning model augmented with more sophisticated components, namely Spatial Pyramid Pooling (SPP), Path Aggregation Network (PANet), and Convolutional Block Attention Module (CBAM) to enhance overall stability and generalization for clinical use. The key components of the proposed model are first trained on the VinBigData Chest X-ray Abnormalities Detection dataset to improve feature extraction and adaptively stretch multiple image resolutions across different strides per pixel levels while focusing more attention on the region. Next, we make a multistage fine-tuning approach for real-world clinical data, which usually shifts domains in practical settings. Finally, the model is forced to be more resistant and less overfitting by performing realworld data augmentation instead of focusing on clinical variability. We further qualitatively assess the performance of our model on both the VinBigData test set and CheXDet dataset with only publicly available bounding box annotations on matching classes between the two datasets. Moreover, the model was integrated into a web application that could easily be employed in clinical environments for real-time chest X-ray analysis and may assist with more accurate diagnosis at an early stage.
Publication Details
Page(s) 1269-1289
DOI DOI not available
Published Journal: International Journal of Communication Networks and Information Security, Volume: 16, Issue: 4, Year: 2024

Request the full text from PASTIC

Keywords
deep learning, Multi-Stage Fine-Tuning, yolov5, Spatial Pyramid Pooling, Path Aggregation Network, Clinical Deployment, Convolutional Block Attention Module, Real-World Data Augmentation, Chest X-ray Abnormalities, Domain Shift
References
VinBigData Chest X-ray Abnormalities,Detection Abnormalities .2020 .. , : .
.2021 .. Imaging Systems and Technology, 31(2) : 1050-1064.
.2020 .". arXiv preprint arXiv:1412.6980, 106 : 6.
.2022 .. , 12 : 1330-1345.
.2024 .models in clinical decision-making." Nature medicine (. , : 1-10.
.2023 .. augmentation strategies." Artificial Intelligence Review, 56(3) : 2111-2169.
.2021 .classification on chest x-ray using weighted multiscale similarity." IEEE Access 9 (. , : 113315.
imagery. .2020 .. Remote Sensing, 13 : 2136.
Xiong Tarek,Zayed Tarek . ."A novel YOLOv8-GAM-. , : .
.2024 .. Materials, 414 : 135025.
.2020 .. industrial big data." Future Internet, 12(6) : 102.
. .Interoperability Using HL7 and DICOM.". In Smart Healthcare and Machine Learning, : 299.
.2024 .. , : .
System Health .2024 .. , : 12-37.
Encryption Robust .2024 .. 5s, 20 : 1993-2000.
.2023 .. trends." Cyber Security and Applications, 1 : 100016.
.2019 .interpretability: A survey on methods and metrics. " Electronics, 8(8) : 832.
.2020 .computer vision and pattern recognition. , : 12595-12604.
0

Citations

0

Downloads

30

Views