DeepLOC: Deep Learning-based Bone Pathology Localization and Classification in Wrist X-ray Images

Razan Dibo,Andrey Galichin,Pavel Astashev,Dmitry V. Dylov,Oleg Y. Rogov

2023-08-24

Abstract:In recent years, computer-aided diagnosis systems have shown great potential in assisting radiologists with accurate and efficient medical image analysis. This paper presents a novel approach for bone pathology localization and classification in wrist X-ray images using a combination of YOLO (You Only Look Once) and the Shifted Window Transformer (Swin) with a newly proposed block. The proposed methodology addresses two critical challenges in wrist X-ray analysis: accurate localization of bone pathologies and precise classification of abnormalities. The YOLO framework is employed to detect and localize bone pathologies, leveraging its real-time object detection capabilities. Additionally, the Swin, a transformer-based module, is utilized to extract contextual information from the localized regions of interest (ROIs) for accurate classification.

Computer Vision and Pattern Recognition,Artificial Intelligence

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to improve the accurate localization and classification of bone pathologies in wrist X - ray images. Specifically, the paper proposes a deep - learning - based method - DeepLOC, which aims to improve the localization and classification of bone pathologies in wrist X - ray images by combining YOLO (You Only Look Once) and Shifted Window Transformer (Swin). This method mainly targets two key challenges:

1. **Accurate bone pathology localization**: Use the YOLO framework to detect and localize bone pathologies, taking advantage of its real - time object detection capabilities.
2. **Accurate anomaly classification**: Use the Swin Transformer module to extract contextual information from the localized regions of interest (ROIs) to achieve accurate classification.

By introducing multi - scale feature fusion and attention mechanisms, the paper improves the classification performance, can effectively capture local and global contextual information, thereby achieving more comprehensive representation learning and improving classification accuracy. Experimental results show that this method significantly improves the accuracy of bone pathology localization and classification compared to existing methods.