New method boosts whole-heart segmentation accuracy across medical imaging sites

Researchers have developed a new method to improve whole-heart segmentation in medical imaging, specifically for computed tomography (CT) and magnetic resonance imaging (MRI) scans. This technique addresses the challenge of generalizing segmentation models across different data acquisition sites and modalities, which often leads to performance degradation. The proposed pipeline combines the TotalSegmentator and nnU-Netv2 models with a novel appearance augmentation strategy that preserves labels while adapting to site-specific characteristics. This approach significantly enhances the Dice scores for CT segmentation, improving it from 0.8350 to 0.9135, and also shows gains in MRI segmentation, from 0.7695 to 0.7830, while reducing the Hausdorff distance 95th percentile (HD95). AI

IMPACT Improves accuracy and robustness of medical image segmentation, potentially leading to better diagnostic tools and treatment planning.

RANK_REASON The item is an academic paper detailing a new method for medical image segmentation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• computed tomography
• Counselor Aligned Response Engine
• magnetic resonance imaging
• nnU-Netv2
• TotalSegmentator

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →