Advancing Ki67 hotspot detection in breast cancer: a comparative analysis of automated digital image analysis algorithms

• Published in: Histopathology Vol. 86; no. 2; pp. 204 - 213
• Main Authors: Zwager, Mieke C, Yu, Shibo, Buikema, Henk J, Bock, Geertruida H, Ramsing, Thomas W, Thagaard, Jeppe, Koopman, Timco, Vegt, Bert
• Format: Journal Article
• Language: English
• Published: England Wiley Subscription Services, Inc 01.01.2025; John Wiley and Sons Inc
• Subjects: Algorithms; artificial intelligence (AI); Automation; Biomarkers, Tumor - analysis; breast cancer; Breast carcinoma; Breast Neoplasms - diagnosis; Breast Neoplasms - metabolism; Breast Neoplasms - pathology; Comparative analysis; Deep Learning; digital image analysis (DIA); Female; hotspot; Humans; Image processing; Image Processing, Computer-Assisted - methods; Immunohistochemistry - methods; Ki-67 Antigen - analysis; Ki-67 Antigen - metabolism; Ki67 proliferation index; Original; Research hot spots; Ki67 proliferation index; breast cancer; hotspot; artificial intelligence (AI); digital image analysis (DIA)
• ISSN: 0309-0167; 1365-2559; 1365-2559
• DOI: 10.1111/his.15294

Aim Manual detection and scoring of Ki67 hotspots is difficult and prone to variability, limiting its clinical utility. Automated hotspot detection and scoring by digital image analysis (DIA) could improve the assessment of the Ki67 hotspot proliferation index (PI). This study compared the clinical performance of Ki67 hotspot detection and scoring DIA algorithms based on virtual dual staining (VDS) and deep learning (DL) with manual Ki67 hotspot PI assessment. Methods Tissue sections of 135 consecutive invasive breast carcinomas were immunohistochemically stained for Ki67. Two DIA algorithms, based on VDS and DL, automatically determined the Ki67 hotspot PI. For manual assessment; two independent observers detected hotspots and calculated scores using a validated scoring protocol. Results Automated hotspot detection and assessment by VDS and DL could be performed in 73% and 100% of the cases, respectively. Automated hotspot detection by VDS and DL led to higher Ki67 hotspot PIs (mean 39.6% and 38.3%, respectively) compared to manual consensus Ki67 PIs (mean 28.8%). Comparing manual consensus Ki67 PIs with VDS Ki67 PIs revealed substantial correlation (r = 0.90), while manual consensus versus DL Ki67 PIs demonstrated high correlation (r = 0.95). Conclusion Automated Ki67 hotspot detection and analysis correlated strongly with manual Ki67 assessment and provided higher PIs compared to manual assessment. The DL‐based algorithm outperformed the VDS‐based algorithm in clinical applicability, because it did not depend on virtual alignment of slides and correlated stronger with manual scores. Use of a DL‐based algorithm may allow clearer Ki67 PI cutoff values, thereby improving the clinical usability of Ki67. Manual assessment of Ki67 hotspots is difficult and prone to variability. Automated Ki67 hotspot assessment correlated strongly with manual Ki67 assessment and provided higher Ki67 PIs. DL‐based automated Ki67 assessment has high clinical applicability, because it does not depend on virtual alignment of slides and correlates strongly with manual scoring.