Skip to content
  • Home
  • Services
    • All Services
    • Custom DICOM Viewer
    • Case Studies
  • Products
    • All Products
    • PYCAD DICOM Viewer
    • PYCAD Dental Viewer
    • PYCAD SSR Engine
  • Case Studies
  • Blog
  • About us
  • Home
  • Services
    • All Services
    • Custom DICOM Viewer
    • Case Studies
  • Products
    • All Products
    • PYCAD DICOM Viewer
    • PYCAD Dental Viewer
    • PYCAD SSR Engine
  • Case Studies
  • Blog
  • About us
Get in Touch

Category: Medical Imaging

Plausible but Unimplantable: Foundation Models vs Cranial Implants

Figure 2 of arXiv:2610.30035 by Wodzinski, Kwarciak, Hemmerling and Daniol (AGH University of Krakow): a real craniectomy skull from MUG500+ with the expert target implant and a task-specific model beside the outputs of MSN-FM, AdaPoinTr and MGPC, each marked not implantable, plus Fig. 1b showing 0 of 239 implantable outputs per model.

An audit of three released shape completion models on 239 skull defects found no output that could be placed in a patient.

SGAV-MP: Smaller Segmentation Models That Keep Their Uncertainty

Research card from arXiv:2610.25460 by Kumar, Sharma, Baruah and Rathore (Indian Institute of Science): Fig. 1 plot of compression ratio against uncertainty-map correlation on MosMedPlus CT, Table 1 showing uniform INT4 keeping Dice while uncertainty correlation drops, and Algorithm 1 of SGAV-MP.

A post-training mixed-precision method that checks calibration and uncertainty maps, not only Dice, before it quantizes a layer.

DART: One Shape Model That Repairs Broken Segmentation Masks

Figures 1 and 2 of the DART paper: 3D renderings of multi-class CT segmentations, example anatomical text descriptions, and the inference row where an imperfect colon prediction from any segmentation tool is cropped, partially noised and passed through a frozen denoising network to a refined mask. Swarna and colleagues, Johns Hopkins University, Aalto University and ELLIS Institute Finland, arXiv:2610.27605

A text-conditioned 3D diffusion model cleans up masks from TotalSegmentator or MedSAM, on CT and MRI, without reading the image.

Hippocampal Segmentation Ensembles That Flag Their Own Failures

Fig. 1 of the paper: sample variance, mutual information and predictive entropy maps from an InnerEye, FastSurfer and HippoSeg ensemble overlaid on T1 MRI around the hippocampus, for a high-Dice example (top) and a low-Dice example (bottom), with the reference boundary in dark red. Oliveira-Stahl and colleagues, UCL and King's College London, arXiv:2610.31505

Averaging InnerEye, FastSurfer and HippoSeg gives calibrated hippocampus masks plus an entropy score that sends doubtful cases to review.

MedBOE: Rebuilding the Full Body Surface From Truncated PET/CT

Four panels from Fig. 1 of the MedBOE paper: a coronal whole-body image, the incomplete body surface extracted from the scan with virtual-marker colours, the initial complete model with estimated arms and hands in red, and the final complete grey body model. Hoijoon Jung, Yuhao Wei and colleagues, University of Sydney, arXiv:2610.28985

A training-free SMPL fit fills in arms and hands cut off by the CT field of view, so body size metrics stop coming out short.

QICOM: Image Quality, Blinded Reads and Radiomics in One App

Method-strip banner of three QICOM application screenshots from the paper: the quantitative workbench comparing a reference and a comparison CT series, the blinded reading view with neutral A and B panels and 1 to 5 score questions, and the radiomics workbench with the same lung VOI in blue on both series. All images show a synthetic thorax phantom. Fabio Mattiussi and Switinder Ghotra, arXiv:2610.31218

Open-source desktop tool that keeps metrics, reader scores and radiomic features tied to the same images. Apache 2.0.

CAP-DIR: Adaptive Radiotherapy Registration That Reads the Plan

Magazine before/after grid from Fig. 3e: coronal treatment CT slices of the head and neck, abdomen and pelvis, with organ-at-risk contours in the top row and target-volume contours in the bottom row. Red is the clinician reference, green is the planning contour copied over without registration, blue is the CAP-DIR propagated contour. Caiwen Jiang, ..., Dinggang Shen, Wei Liu, Mayo Clinic, arXiv:2610.18926

Uses planning contours, dose, case notes and earlier fractions to carry contours onto repeat CTs. No code yet; weights stay private.

ReSeg: Organ Masks on CT, MRI and PET Without Real Training Scans

Magazine before/after grid from Figs. 8 and 9: an abdominal CT slice from AMOS-CT and an abdominal MRI slice from AMOS-MR, each shown as the input, the reference labels and the ReSeg prediction, with liver in red, spleen in green, right kidney in blue and left kidney in yellow. Derong Yu, Guoyan Zheng, Shanghai Jiao Tong University, arXiv:2610.29090

Trained only on synthetic volumes built from organ masks, it outlines liver, spleen and kidneys on CT, MRI and PET. MIT code, no weights.

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

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

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

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

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

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.

← Previous
Next →