v34.0-adaptive · Checkpoint 652

I-Translation

Bidirectional CT ↔ MRI medical image translation powered by 4 independent U-Net generators

CT Scan
GI
CT→MRI
MRI→CT
MRI Scan
FJ

Fidelity Notice

Match input modality to the correct block. Images are auto pre-processed to 64×64 grayscale before AI inference. For DICOM series, select the best slice before converting — only the active slice is sent to the model.

CT → MRI Conversion

Generator G · Upload a CT scan

GI CT→MRI

Drop CT scan here

PNG, JPG, TIFF, or .DCM (DICOM) — single file or multi-slice series

Auto-conversion active — PNG/JPG/TIFF images are auto-resized to 217×181. DICOM files are auto-converted to PNG at 217×181 before sending to the AI model.

MRI → CT Conversion

Generator F · Upload an MRI scan

FJ MRI→CT

Drop MRI scan here

PNG, JPG, TIFF, or .DCM (DICOM) — single file or multi-slice series

Auto-conversion active — PNG/JPG/TIFF images are auto-resized to 217×181. DICOM files are auto-converted to PNG at 217×181 before sending to the AI model.

Generator Reference

Generator Direction Input Output API type
GCT→MRICT ScanMRI Outputct_to_mriPrimary Convertor (70%)
ICT→MRICT ScanMRI Fidelityct_to_mriFidelity Enhancer (30%)
FMRI→CTMRI ScanCT Outputmri_to_ctPrimary Convertor (70%)
JMRI→CTMRI ScanCT Fidelitymri_to_ctFidelity Enhancer (30%)
G+I fidelityCT→MRICT ScanMRI Outputct_to_mriSingle Output · G(70%)+I(30%) internal blend
F+J fidelityMRI→CTMRI ScanCT Outputmri_to_ctSingle Output · F(70%)+J(30%) internal blend