Papers

AIstatLab 提出 BraTS-GoAT 2026 Task 3 脑肿瘤分割方案,验证集 mean DSC 达 0.834

AIstatLab at BraTS-GoAT 2026 Task 3: Quality-Controlled Consensus Pseudo-Labeling and Dense Expert Adaptation

Why we picked it

做医学影像分割的可以看看:他们用五折教师模型给 1138 例无标注数据打伪标签,筛掉不合格的后混进 1351 例真实标注一起训练,代码也开源了。ET 区域提升最明显,消融写得挺诚实,各组件效果其实不大。

AIstatLab 团队发布了 BraTS-GoAT 2026 Task 3(跨异质性脑肿瘤实体分割)的技术方案。最终提交的五折 ResEnc-M 模型在验证服务器上取得 mean DSC 0.834、mean NSD 0.510、Acalc = 0.6720。

Body

方案核心是一个 3D ResidualEncoderUNet,训练时使用了质量控制的一致性伪标签:五折 ResEnc-M/L 教师模型先对 1,138 例无标注病例做预测,经区域层级约束和置信度-一致性双重筛选后保留 983 例,与 1,351 例已标注数据合并训练。损失函数仅在训练阶段使用嵌套区域目标,模型内含一个四专家瓶颈适配器(four-expert bottleneck adapter)。

消融结果显示各组件的收益有限且随区域变化,其中在 ET(增强肿瘤)区域上的提升最明显。源代码已在 GitHub 开源。

Original · arXiv cs.AI

AIstatLab at BraTS-GoAT 2026 Task 3: Quality-Controlled Consensus Pseudo-Labeling and Dense Expert Adaptation

BraTS-GoAT Task 3 evaluates brain-tumor segmentation across heterogeneous entities using shared ET, TC, and WT regions. We present a 3D ResidualEncoderUNet trained with quality-controlled consensus pseudo-labels, a training-only nested-region objective, and a four-expert bottleneck adapter. Five-fold ResEnc-M/L teachers gener ated predictions for 1,138 unlabeled cases; fusion, hierarchy enforcement, and confidence-and-agreement screening retained 983 for training with 1,351 annotated cases. The final five-fold ResEnc-M entry achieved mean DSC 0.834, mean NSD 0.510, and Acalc = 0.6720 on the validation server. Component effects were modest and region-dependent, with the clearest gains on ET. Source code is available at https://github.com/HeeSeung-Shin/Brats2026-Goat-HS.

Industry Coordinates & Source Verification

This report is categorized under AITOP's "Papers" desk. Within our cross-regional analytical framework, it directly maps to AI developer toolchains, coding agents, and engineering workflows. Key entities and technical topics under continuous tracking include BraTS-GoAT, ResEnc-M, medical imaging, image segmentation.

Source attribution: curated from arXiv cs.AI (@Heeseung Shin, Changwon Lim).