Chest X-ray Pneumothorax Segmentation Using EfficientNet-B4 Transfer Learning in a U-Net Architecture
This addresses pneumothorax detection for radiologists, but it is incremental as it applies a known hybrid method to a specific medical imaging task.
The paper tackled automated segmentation of pneumothorax in chest X-rays using a U-Net with EfficientNet-B4 encoder, achieving an IoU of 0.7008 and Dice score of 0.8241 on an independent dataset.
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.