Guides, project breakdowns, and lessons learned from building medical imaging platforms — written by the PYCAD team.
nnU-Net proposes femoral, tibial, and patellar cartilage on 3D knee MRI; two readers correct, a third adjudicates, then volume, 3D-RT thickness, and 3D-RBA thin-area run on the frozen mask. Yang et al., arXiv:2609.08081.
CISS MRI plus TOF-MRA for trigeminal planning: Zhang et al. show whole-brain registration scores mislead. ROI-centered, vessel-aware, contrast- and FOV-aware checks on 149 MVD patients. arXiv:2609.08805.
TAS-Net-D extends TAS-Net with persistent memory tokens and teacher-student fine-tuning so echo myocardial strain curves close nearer end-diastole across full cycles. Judge et al., arXiv:2609.09577.
Coarse COVID report text plus clinician-scored radiology concepts steer CXR and CT lesion overlays in CGSM. Lin et al., arXiv:2609.07004.
Sybil boxes, NLSTseg lesions, and TotalSegmentator organs on NLST screening CT, shipped as DICOM SEG and Structured Reports you can hang in OHIF or Slicer. Krishnaswamy et al., arXiv:2609.10858.
We build custom medical imaging platforms — advanced DICOM viewers, AI segmentation, and the clinical systems around them.
Copyright © 2026 PYCAD. All Rights Reserved.