A local tool for liver tumour volumetry. Load a CT, contour the lesion slice by slice, see it in 3D, put the follow-up scan alongside the baseline, and export a report for the MDT — without the images ever leaving the machine.
Demo on a public, de-identified case: HCC_013 from the HCC-TACE-Seg collection — Moawad et al. (2021), The Cancer Imaging Archive, doi:10.7937/TCIA.5FNA-0924, CC BY 4.0. Pre-TACE contours were traced from the dataset’s expert segmentation; LiverVol measured 110.9 cm³ against the 108.5 mL reference.
Liver lesions are irregular, and the one- or two-axis measurements used day to day can move a long way from the true tumour burden. For treatment planning — and for judging response after TACE or ablation — the number that matters is volume, but measuring it usually means specialist software, a workstation, and time no one has between lists.
LiverVol is a local desktop tool: a small Python server and a browser interface, started with one script. Drop in a DICOM series and the viewer is ready. Lesions are contoured with semi-automatic tools — click a seed point, drag a box, propagate to neighbouring slices — or traced freehand, then refined with paint, erase and keep brushes. Volume is computed from every contoured slice, a per-slice area chart shows the lesion's profile, a 3D mesh is built on demand, and a follow-up study can be loaded alongside the baseline for side-by-side review.
| Backend | Python + FastAPI, pydicom, NumPy, scikit-image, SciPy |
| Frontend | React + TypeScript, canvas-based DICOM viewer, three.js 3D view |
| Reports | ReportLab PDF generation |
| Runs | Locally, started with one script; no cloud services |
A working local tool. Next on the list: a growth limit for slice-to-slice propagation, which can spread into normal liver on low-contrast lesions, and automatic liver segmentation as an anatomical overlay.
If your department measures by eye because the proper tools are out of reach, there's usually a simpler way to get the real number.