Chest X-ray Pneumothorax Segmentation Using EfficientNet-B4 Transfer Learning in a U-Net Architecture

2025/09/04 by Roque, Alvaro Aranibar, Sebastian, Helga
#Artificial Intelligence (cs.AI) #Computer Vision and Pattern Recognition (cs.CV) #FOS: Computer and information sciences

paper · doi:10.48550/arxiv.2509.03950

Abstract

Pneumothorax, the abnormal accumulation of air in the pleural space, can be life-threatening if undetected. Chest X-rays are the first-line diagnostic tool, but small cases may be subtle. We propose an automated deep-learning pipeline using a U-Net with an EfficientNet-B4 encoder to segment pneumothorax regions. Trained on the SIIM-ACR dataset with data augmentation and a combined binary cross-entropy plus Dice loss, the model achieved an IoU of 0.7008 and Dice score of 0.8241 on the independent PTX-498 dataset. These results demonstrate that the model can accurately localize pneumothoraces and support radiologists.

Citations

• EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
• Albumentations: Fast and Flexible Image Augmentations
• Revisiting Small Batch Training for Deep Neural Networks
• U-Net: Convolutional Networks for Biomedical Image Segmentation

Related

• Pneumothorax Segmentation: Deep Learning Image Segmentation to predict Pneumothorax
• The 2ST-UNet for Pneumothorax Segmentation in Chest X-Rays using ResNet34 as a Backbone for U-Net
• Deep Learning for Pneumothorax Detection and Localization in Chest Radiographs
• X-Recon: Learning-based Patient-specific High-Resolution CT Reconstruction from Orthogonal X-Ray Images
• Leveraging Anatomical Constraints with Uncertainty for Pneumothorax Segmentation