Learning from the few: Fine-grained approach to pediatric wrist pathology recognition on a limited dataset

TL;DR

This work reframes pediatric wrist pathology recognition as a fine-grained visual recognition task to overcome data scarcity in X-ray analysis. It introduces a Plug-in Module for FGVR (PIM) with a Swin Transformer backbone, weakly supervised region selection, graph-based feature fusion, and Grad-CAM explainability, augmented by the LION optimizer for better generalization. Through careful dataset curation from GRAZPEDWRI and multiple augmentation schemes, the approach achieves higher accuracy than state-of-the-art baselines, with notable gains when combining augmentation and LION. The study also demonstrates robust interpretability via heatmaps and discusses limitations such as the need for external validation and larger datasets, outlining clear directions for future work and potential clinical impact in improving pediatric wrist injury diagnosis with limited data.

Abstract

Wrist pathologies, {particularly fractures common among children and adolescents}, present a critical diagnostic challenge. While X-ray imaging remains a prevalent diagnostic tool, the increasing misinterpretation rates highlight the need for more accurate analysis, especially considering the lack of specialized training among many surgeons and physicians. Recent advancements in deep convolutional neural networks offer promise in automating pathology detection in trauma X-rays. However, distinguishing subtle variations between {pediatric} wrist pathologies in X-rays remains challenging. Traditional manual annotation, though effective, is laborious, costly, and requires specialized expertise. {In this paper, we address the challenge of pediatric wrist pathology recognition with a fine-grained approach, aimed at automatically identifying discriminative regions in X-rays without manual intervention. We refine our fine-grained architecture through ablation analysis and the integration of LION.} Leveraging Grad-CAM, an explainable AI technique, we highlight these regions. Despite using limited data, reflective of real-world medical study constraints, our method consistently outperforms state-of-the-art image recognition models on both augmented and original (challenging) test sets. {Our proposed refined architecture achieves an increase in accuracy of 1.06% and 1.25% compared to the baseline method, resulting in accuracies of 86% and 84%, respectively. Moreover, our approach demonstrates the highest fracture sensitivity of 97%, highlighting its potential to enhance wrist pathology recognition. The implementation code can be found at https://github.com/ammarlodhi255/fine-grained-approach-to-wrist-pathology-recognition

Figures (8)

• Figure 1: Two species of gulls from the CUB200 dataset, illustrating the difficulty of fine-grained recognition.
• Figure 2: Large intra-class variance within fracture pathology.
• Figure 3: Small inter-class variance between three different pathologies: fracture (left), bone anomaly (middle), and soft tissue (right), respectively.
• Figure 4: Proposed methodology depicting dataset curation, model training and evaluation, network refinement, and interpretability.
• Figure 5: Illustration of dataset curation steps.