Automatic deep learning-based tumor segmentation for PET-CT in mice with an uncertainty quantification module

Nature
Read full story· 5 min
Researchers developed TumSeg, a deep learning model that automatically segments subcutaneous tumors in mice CT scans with high accuracy and estimates tumor volume and PET scan metrics. Validated on 452 scans, TumSeg outperforms prior methods and includes an uncertainty module to assess segmentation quality. It also adapts well to MRI scans, enhancing preclinical tumor analysis efficiency and reproducibility.


More in Healthcare