PYCAD BLOG

Insights on Medical Imaging, AI & DICOM

Guides, project breakdowns, and lessons learned from building medical imaging platforms — written by the PYCAD team.

Magazine before/after from Fig. 3: the same portable AP chest radiograph of a child in a cardiac ICU, on the left with sternal wires, tubes, lines and ECG leads, on the right after PCXR-Copilot masked the devices and filled them in with LaMa inpainting. Li, Girten, Zeng, Jiang, Huang, Xu, Chen, Shi, Kheir, arXiv:2610.25486

PCXR-Copilot: Pediatric ICU Chest X-Rays Without the Device Clutter

Masks tubes and wires, paints them out, then scores lung water and body-wall edema. Live at Boston Children’s. Code not out yet.

October 21, 2026

Magazine before/after from Fig. 1: one total-body FDG PET scan reconstructed from 10% of counts by the vendor, with smoothing, with Image-DDS and with TB-PET-DDS next to the full-dose reference, coronal view and heart zoom with slice NRMSE numbers. Dassanayake, Webber, Prokopenko, Schnabel, Reader, arXiv:2610.23870

TB-PET-DDS: A Diffusion Prior for Low-Count Total-Body PET

A CT-conditioned diffusion prior plus a list-mode PET data step, tested on one total-body FDG scan at 10% counts. No code yet.

October 21, 2026

Magazine before/after from Figs. 4 and 5 of NV-Generate-Body: coronal, sagittal and axial views of generated 256-cube CT label maps, top row liver requested at the 20th and 80th percentile with the same seed, bottom row BMI slot set to 18 and 40. Zhao, Guo, ..., Turkbey, Xu, NVIDIA, NIH, UAMS, arXiv:2610.24153

NV-Generate-Body: Synthetic Whole-Body CT Masks You Can Dial In

NVIDIA’s NV-Generate-Body draws whole-body CT label maps with the organ sizes, tumors, sex and BMI you ask for. Code Apache 2.0, weights public.

October 20, 2026

Magazine before/after from Fig. A5: coronal contrast-enhanced T1 of one brain-metastasis case rebuilt from 3 mm and 4 mm slices by cubic interpolation, SAINT and GSMR3D next to the real 1 mm scan, with detected (red) and missed (blue) lesion boxes. Rastogi, Leiser, ..., Brugnara, Vollmuth, arXiv:2610.24858

GSMR3D: One Model Turns Thick-Slice Brain MRI Into 1 mm Volumes

GSMR3D rebuilds thick-slice brain MRI on a 1 mm grid at any spacing, one model per contrast, and kept small metastases detectable.

October 20, 2026

Magazine before/after from Fig. 4: two held-out brain tumor patients at 16x and 12x acceleration, each shown as the prior visit used as guide, SENSE, unguided LPDSNet, LGGS and the fully sampled scan, with zoom and error strips. Janjusevic, Chandarana, Wang, Feng, arXiv:2610.22643

LGGS: Faster Follow-Up Brain MRI Using the Patient’s Last Scan

LGGS rebuilds heavily undersampled follow-up brain MRI using the patient’s prior exam as a guide, without copying it into the new image. arXiv:2610.22643.

October 19, 2026

Magazine before/after from Fig. 3: two brain MRI test cases, a 64 mT FLAIR and a 3T T1, each shown as the lower-field input, the UniField++ virtual higher-field output and the real 3T or 7T scan. Lin, Wang, Yao, Li, Sun, Deng, Yuan, arXiv:2610.22982

UniField++: Making Low-Field Brain MRI Look Like 3T and 7T

UniField++ turns 64 mT brain MRI into virtual 3T and 3T into virtual 7T, with a vision-language planner that flags regions likely to go wrong. arXiv:2610.22982.

October 19, 2026

We build custom medical imaging platforms — advanced DICOM viewers, AI segmentation, and the clinical systems around them.