arXiv：cs.LG· Minh Sao Khue Luu, Evgeniy N. Pavlovskiy, Bair N. Tuchinov·· 6 小时前AI 评分36
CATMIL：面向脑MRI小病灶分割的组件自适应与病灶级监督
Component-Adaptive and Lesion-Level Supervision for Improved Small Structure Segmentation in Brain MRI
AI 导读
CATMIL 在不改架构的前提下为 nnU-Net 的 Dice 与交叉熵损失加入两项辅助项，在多发性硬化病灶分割数据集 MSLesSeg 上将小病灶召回率提升至 0.873（Dice+CE 为 0.796），漏检减少约 48%，DSC 与 HD95 相当，对直径≥3 mm 的临床读片尺寸病灶召回率达 0.944。
正文
Abstract:Small lesions in brain MRI are hard to segment because they occupy a tiny fraction of the volume and are dominated by background and larger lesions during voxel-wise optimization, so a model can reach a high Dice similarity coefficient (DSC) while missing many of them. We propose CATMIL, a training objective that adds two auxiliary terms to the standard nnU-Net Dice and cross-entropy loss without changing the architecture. The Component-Adaptive Tversky (CAT) term weights lesion voxels by the inverse size of their connected component, so each lesion contributes nearly equally regardless of volume. The lesion-level Multiple Instance Learning (MIL) term treats each lesion as a bag of voxels and penalizes lesions with no detected voxel. For multiple sclerosis lesion segmentation on MSLesSeg, CATMIL achieves the highest small-lesion recall (0.873 vs. 0.796 for Dice+CE; 95% CI of the difference +0.030 to +0.157, higher in all six test patients) and about 48% fewer missed lesions, with comparable DSC and HD95. The gain holds for lesions of at least 3 mm in diameter, the clinical reading size (recall 0.944 vs. 0.870). Standard losses produce no probability response to most small lesions they miss, so no threshold can recover them. The cost is more small false-positive components; a simple component-size filter removes most of them while keeping the sensitivity gain, and at matched lesion-wise precision CATMIL detects more small lesions with higher lesion-wise F1. An ablation attributes the detection gain to the MIL term. On a second dataset, 3D-MR-MS, CATMIL with the same loss weights again improves small-lesion recall, at a larger false-positive cost and slightly lower DSC. Code: this https URL
| Comments: | This version added evaluation on a second dataset (3D-MR-MS) and a held-out test set; added statistical significance tests and error analysis; added new references; corrected the optimizer description; update figures |
| Subjects: | Computer Vision and Pattern Recognition (cs.CV); Machine Learning (cs.LG) |
| Cite as: | arXiv:2604.08015 [cs.CV] |
| (or arXiv:2604.08015v3 [cs.CV] for this version) | |
| https://doi.org/10.48550/arXiv.2604.08015 arXiv-issued DOI via DataCite |
Submission history
From: Minh Sao Khue Luu [view email]
[v1]
Thu, 9 Apr 2026 09:15:10 UTC (2,873 KB)
[v2]
Mon, 27 Apr 2026 11:00:26 UTC (3,326 KB)
[v3]
Wed, 7 Oct 2026 14:27:39 UTC (378 KB)
来源：arXiv：cs.LG · arxiv.org