Pneumothorax Segmentation Using Low-Cost Computing Method

Abstract

Pneumothorax, a life-threatening condition characterized by air leakage in the pleural space, presents a significant diagnostic challenge in chest X-rays (CXR). Deep learning methods have shown promise in medical image segmentation, but they face limitations due to data scarcity and computational demands. This study investigates the applicability of traditional segmentation techniques, including Uniform Local Binary Patterns (ULBP), filtering, clustering, and manual pattern selection, for pneumothorax segmentation in CXR images. Our results demonstrate that these traditional methods yield significantly lower Dice coefficient scores (below 0.20) compared to the desired threshold, indicating their limitations in accurately segmenting pneumothorax. These findings highlight the challenges associated with traditional approaches for pneumothorax segmentation and emphasize the need for more advanced techniques, such as deep learning, to address the complexities of this medical imaging task.