Researchers develop ProSeg for diverse and personalized medical image segmentation

Researchers have developed ProSeg, a novel probabilistic modeling approach for multi-rater medical image segmentation. This method addresses the challenge of inter-observer variability and ambiguous lesion boundaries by introducing latent variables to capture expert preferences and boundary uncertainty. ProSeg allows for the generation of segmentation outputs that are both diverse and personalized to individual annotators, achieving state-of-the-art performance on nasopharyngeal carcinoma and lung nodule datasets. AI

影响 Introduces a new method for improving the accuracy and interpretability of medical image segmentation by accounting for annotator variability.

排序理由 This is a research paper detailing a new probabilistic modeling approach for medical image segmentation. [lever_c_demoted from research: ic=1 ai=1.0]

在 arXiv cs.CV 阅读 →

• Ke Liu
• LIDC-IDRI

• 论文
• 其他

AI 生成摘要 · Google Gemini · 来自 1 个来源。 我们如何撰写摘要 →