Every liver lesion, followed from one scan to the next.
Scanvero finds and measures liver metastases on each CT scan, links them across a patient's follow-up scans, and reports what changed. The radiologist confirms every link before a number appears.
Research software from Scanvero Ltd. Not for clinical use.
Slices from the Scanvero research build on public research CT (CRLM-CT-1006). Not patient care data.
The same patient, scanned every few months.
When colorectal cancer spreads to the liver, patients are re-scanned every 2 to 3 months through chemotherapy, ablation and surgical planning. Each time, someone has to find every lesion again and work out which is which.
SlowSeveral lesions, several scans, all re-measured by hand at each visit.
InconsistentTwo readers measuring the same lesion often get different answers.
ConsequentialThe size change decides whether treatment is working.
Why it matters.
More patients, more scans, fewer radiologists, and measurements that vary between readers. Each figure links to its original source.
developed liver metastases within five years, in a population study of 1,026 patients.
Engstrand et al., BMC Cancer 2018 (Stockholm)how far two radiologists' measured change in one lesion can differ, wider than the 20% growth that counts as progression.
Yoon et al., Eur J Cancer 2016, 12 studiesfewer radiologists than the UK needs today, forecast to reach 40% by 2030.
Royal College of Radiologists, 2025 censusNo confirmation, no number.
Scanvero would rather show "not available" with a reason than a plausible number nobody has checked. Try it: the change in size appears only once the link is confirmed.
- Contrast phase not confirmed: measurement withheld.
- Lesion link not confirmed: change withheld.
- Result cannot be traced to its source images: result withheld.
Lesion 2, segment VII
Illustration with example valuesA demonstration of the rule, not output from a patient scan.
What we have measured so far.
Research results on public data, with pass marks written down before each test. Read the full evidence and its limits.
Liver outline overlap
Average Dice against reference outlines, with 1.17% average volume error, on 102 of 103 external colorectal cases (MCT-LTDiag, 1 excluded).
Lesions over 20 mm found
And 78.5% of lesions 10 to 20 mm, on 103 external colorectal cases (MCT-LTDiag). False positives, 2.26 per scan, are still above target.
Lesion links correct
Expert outlines, in the 37 of 43 sealed scan pairs the alignment accepted; the other 6 were refused, not guessed. Public melanoma longitudinal research set.
These are research results on public data, not clinical validation.
Work with us
We are looking for hospitals and research groups with baseline and follow-up CT of colorectal liver metastases, radiologists for a reader study, and academic and NHS partners for funded validation.