Authors
Aaishah R Raquib, Steven N Schwartz, Shan H Siddiqi, Hemant Varma
Published in
Journal of neuropathology and experimental neurology. Aug 21, 2026. Epub Aug 21, 2026.
Abstract
Ki-67 is a proliferative biomarker used for diagnosis, grading, and prognosis of various cancer types. This involves manually counting up to 1000 cells, a laborious and error-prone process due to interobserver variability. We developed and validated an open access automated tool for quantification of Ki-67-positive cells in meningiomas, the most common primary brain tumor. The automated counting algorithm was trained using immunohistochemical Ki-67-stained sections from 10 meningioma samples and tested on another 50 meningioma samples. Automated counts were compared to manual counts from 2 independent observers using intraclass correlations (ICC) and Pearson correlation. The 2 observers yielded similar counts across all 50 slides (ICC = 0.802, Pearson r = 0.77). The automated count showed significantly greater correlation with both manual counts (ICC = 0.94, P = 2.5×10-5; ICC = 0.90, P = 0.011). The Pearson correlation between the automated count and both manual counts (r = 0.89 in both cases) was significantly stronger than the correlation between the 2 manual counts (P < 10-6 in each case). Thus, the automated Ki-67 count predicted both manual counts better than manual counts predicted each other. The algorithm is freely available as a web application (https://Ki67.app) and as open-source code (https://github.com/aaishahraquib/Ki67-counter). This freely available counter will provide a user-friendly, time efficient, and accurate resource for practicing pathologists.
PMID:
42627773
Bibliographic data and abstract were imported from PubMed on 22 Aug 2026.
Read full publication at:
Please sign in
to see all details.
Advertisement
Stats
- Recommendations n/a n/a positive of 0 vote(s)
- Views 11
- Comments 0