Precise ablation zone segmentation on CT images after liver cancer ablation using semi-automatic CNN-based segmentation

This paper develops a semi-automatic method to segment the liver ablation zone in CT images after thermal ablation for liver cancer treatment. We utilize 3D nn-UNet to segment the ablation zone coarsely; the incorrectly segmented region is then refined by user click points using RITM — an interactive segmentation method. The proposed method achieves state-of-the-art performance on ablation segmentation accuracy; on average, only 2 minutes are required to correct the segmentation.

Recommended citation: Le Quoc Anh, Xuan Loc Pham, Theo van Walsum, Viet Hang Dao, Tuan Linh Le, Daniel Franklin, Adriaan Moelker, Vu Ha Le, Nguyen Linh Trung, and Manh Ha Luu. "Precise ablation zone segmentation on CT images after liver cancer ablation using semi-automatic CNN-based segmentation." Medical Physics, 51(12):8882–8899, 2024.
Download Paper