Citation

Bannah, Hasanul and Fauzi, Mohammad Faizal Ahmad and Mansor, Sarina and Nabi, Md Serajun and Hossen, Md. Sabbir and Santo, Istiyak Amin and Chiew, Seow Fan and Looi, Lai Meng (2025) Real-Time Nuclei Classification and Allred Scoring in PR-IHC Stained Breast Cancer Histopathology Images. In: 2025 9th International Conference on Information Technology (InCIT), 12-14 November 2025, Phuket, Thailand.

Text 467.pdf - Published Version Restricted to Repository staff only Download (4MB)

Abstract

Progesterone receptor immunohistochemistry (PR-IHC) of breast cancer tissue requires precise nuclei classification and Allred scoring for biomarker quantification and clinical decision-making. However, these tasks are still difficult because of overlap, heterogeneous nuclear morphology, and variations in diaminobenzidine (DAB) staining. Using auto-calibrated DAB-intensity thresholds, we propose a real-time, lightweight framework for classifying individual nuclei into Strong, Moderate, Weak, and Negative categories. The pipeline separates the DAB signal, extracts nuclear contours, aggregates per-nucleus intensity, and uses calibrated, intensity-driven decision-making to assign a single label. By combining the percentage of positive nuclei with their dominant staining intensity, we calculate ROI-level Allred scores, building on the nucleus-level outputs. The University of Malaya Medical Centre (UMMC) provided the primary dataset, and a hybrid automated-manual workflow with expert validation was used to create pixel-based ground truth for 250 PR-IHC images. This approach consistently performs well on a held-out test split (macro-F1 = 0.948; overall IoU = 0.906). The results demonstrate the usefulness of intensitydriven, computationally efficient techniques for Allred scoring and nucleus classification in digital pathology workflows. Instant Allred scoring without sophisticated computation is made possible by the lightweight design, which makes it appropriate for overloaded hospitals and clinics with limited resources.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Progesterone receptor, nuclei classification, Allred score, breast cancer, digital pathology, image process ing, immunohistochemistry
Subjects:	R Medicine > RC Internal medicine > RC0254 Neoplasms. Tumors. Oncology (including Cancer)
Divisions:	Faculty of Artificial Intelligence & Engineering (FAIE)
Depositing User:	Ms Suzilawati Abu Samah
Date Deposited:	19 Mar 2026 00:34
Last Modified:	19 Mar 2026 00:34
URII:	http://shdl.mmu.edu.my/id/eprint/15503

Downloads

Edit (login required)