Premium
Computer‐assisted pathological immunohistochemistry scoring is more time‐effective than conventional scoring, but provides no analytical advantage
Author(s) -
Ong Chee Wee,
Kim Lay Gek,
Kong Hui Hui,
Low Lai Yee,
Wang Ting Ting,
Supriya Srivastava,
Kathiresan Manickam,
Soong Richie,
SaltoTellez Manuel
Publication year - 2010
Publication title -
histopathology
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.626
H-Index - 124
eISSN - 1365-2559
pISSN - 0309-0167
DOI - 10.1111/j.1365-2559.2010.03496.x
Subject(s) - pathological , cytokeratin , medicine , scoring system , pathology , immunohistochemistry , tissue microarray , univariate , computer science , machine learning , multivariate statistics
Ong C W, Kim L G, Kong H H, Low L Y, Wang T T, Supriya S, Kathiresan M, Soong R & Salto‐Tellez M
(2010) Histopathology 56 , 523–529 Computer‐assisted pathological immunohistochemistry scoring is more time‐effective than conventional scoring, but provides no analytical advantageAims: Interpretation of immunohistochemistry is primarily done through human visual scoring while computer‐assisted scoring is relatively uncommon. This study aimed to examine (i) the level of agreement between human visual and computer‐assisted pathological scoring of immunoreactivity expression in colorectal cancers, (ii) whether computer‐assisted scoring affects the prognostic significance of biomarkers, and (iii) whether computer‐assisted pathological scoring provides any time‐saving or reproducibility advantages. Methods and results: Tissue microarray blocks were constructed from the primary colorectal adenocarcinoma specimens of 486 patients. Scoring of the six markers [cytokeratin (CK) 7, CK20, cyclooxygenase‐2, Ki67, p27 and p53] was done independently by a qualified pathologist, a trained scientist and the Ariol SL‐50 (Applied Imaging). Univariate analysis showed that human visual and computer‐assisted scoring were strongly correlated (all κ values >0.8). Both human visual and computer‐assisted pathological scoring identified the same set of biomarkers with significant association with survival. Computer‐assisted pathological scoring was shown to be a time‐effective means of scoring larger numbers of slides (for high‐throughput studies). Conclusions: Our results suggest that computer‐assisted pathological scoring can be a viable alternative to pathologist scoring in a manner that is more practical and time‐effective, but, interestingly, providing no analytical advantage.