Medical Research, Tools, and Ideas

医学研究笔记

Posted by Jing on January 10, 2023 Last updated:

Research collection

A growing collection of papers, code, tools, datasets, and ideas in medical research.

Resource Year Topic What it offers Type Last accessed
MICCAI Open Source Papers 2019 MICCAI 2019-2026 Open Source Papers Collection of MICCAI papers with links to open-source code and dataset information. Paper, Code 2026-10-08
ADA4MIA 2024 Cross-hospital domain adaptation; active learning
Methods for adapting models across hospitals with limited target-domain labels.

医院之间的设备、扫描协议、图像质量和患者分布差异会造成域偏移,使医院 A 训练的模型在医院 B 性能下降;同时,B 可能没有标注或只能负担少量标注。这个仓库收集论文、代码和数据集,研究如何以较低标注成本适应新医院。

  • 无监督领域自适应:利用 B 的无标注数据进行适应,经典设定通常仍可访问 A 的有标注数据。
  • 无源领域自适应:适应时无法访问 A 的原始数据,只使用 A 训练好的模型和 B 的无标注数据。
  • 无源主动领域自适应:在无法访问 A 数据的限制下,主动挑选少量 B 的样本请专家标注,再结合其余无标注数据进行适应。

“Traditional Domain Adaptation” 栏目也包含无源方法,分类并非互斥;这是资源集合,而非单一算法。

Paper, Code, Dataset 2026-10-09
VoxTell 2026 Text-prompted 3D medical image segmentation; CVPR 2026 Another work from the nnU-Net team (MIC-DKFZ): segments anatomical and pathological structures in CT, PET, and MRI using free-text prompts. Paper, Code 2026-10-09
nnInteractive 2025 Interactive 3D medical image segmentation From the same MIC-DKFZ team as nnU-Net and VoxTell: semi-automatic 3D segmentation with points, scribbles, boxes, and lasso prompts; supports iterative refinement. Paper, Code 2026-10-09
TotalSegmentator 2022 CT/MRI anatomical segmentation Built on nnU-Net by University Hospital Basel: automatically segments 117 anatomical classes in CT and 50 in MRI; includes pretrained models and public datasets. Paper, Code, Dataset 2026-10-09
LION 2026 PET; tumor segmentation Automated lesion segmentation for FDG and PSMA PET scans. Paper, Code 2026-10-08
Human Atlas 2026 3D anatomy; medical education Interactive anatomy explorer with 2,234 selectable BodyParts3D meshes, system layers, search, and exploded views. Code, Web Platform 2026-10-09
The Virtual Heart 2024 Cardiac anatomy; medical education Interactive 3D heart visualization and cardiac murmur education. Web Platform 2026-10-08
Resource types
  • Literature: Papers, reviews, preprints, and research reports.
  • Code: Algorithm implementations and reproduction repositories.
  • Tool / Platform: Software, web applications, and analysis platforms.
  • Dataset: Open data, annotations, and benchmark collections.
  • Tutorial / Resource: Courses, technical guides, and educational materials.
  • Research Idea: Questions, project proposals, and collaboration directions.