jst/medics transforms CT DICOM files into interactive 3D anatomical models for surgical planning, simulation, and pre-operative discussion. The platform uses deep-learning AI to automatically segment and differentiate anatomical structures with sub-millimeter precision, delivering clinical-ready models within 48 hours. The service targets renal, kidney stone, and pancreas surgery planning, with each reconstruction priced at €250.
Funding
Funding not disclosed
Founders
Product
Problem
Surgeons and clinical teams often rely on flat, two-dimensional CT images to plan complex procedures, making it difficult to fully appreciate vascular relationships, tumor boundaries, and critical anatomical structures. Traditional 3D reconstruction services are slow, require manual or semi-automatic segmentation, and often cannot differentiate adjacent structures with similar Hounsfield Unit values, limiting their utility in surgical planning.
Solution
jst/medics provides an AI-powered platform that converts CT DICOM files into interactive, surgical-grade 3D anatomical models. The service uses deep-learning neural networks to perform voxel-by-voxel segmentation, automatically identifying and color-coding arteries, veins, parenchyma, tumors, and collecting systems—even when structures have similar radiodensity. Surgeons upload CT scans through a secure web interface, and within 48 hours receive a clinical-ready 3D model accessible through a private dashboard. These models support pre-operative planning, surgical simulation, and educational use, with interactive tools for measuring distances, volumes, and simulating surgical approaches.
Target Audience
Primary customers are radiologists, surgeons, and clinical teams specializing in urology and pancreatic surgery, as well as universities, residency programs, simulation centers, and educational labs seeking 3D reconstructions for training and pre-operative planning.
Features
- Fully automated AI pipeline using deep-learning segmentation for voxel-accurate 3D reconstruction from DICOM scans
- Tissue differentiation beyond HU-based methods, recognizing adjacent structures with similar values including veins, parenchyma, sinus fat, cysts, and tumors
- Color-coded automatic identification of arteries, veins, parenchyma, tumors, and collecting systems
- Interactive surgical planning tools with distance and volume measurements and approach simulation
- 48-hour delivery guarantee from upload to clinical-ready model, without manual segmentation
- Sub-millimeter anatomical fidelity with DICOM-compliant secure ingestion
- Specialized workflows for renal masses, kidney stones (stone burden and HU density mapping), and pancreas tumors (gland segmentation and lesion localization)
- Automatic anonymization of patient data upon upload, with models available on a private dashboard for 30 days