Relevance of deep learning to facilitate the diagnosis of HER2 status in breast cancer.

摘要: Tissue biomarker scoring by pathologists is central to defining the appropriate therapy for patients with cancer. Yet, inter-pathologist variability in interpretation of ambiguous cases can affect diagnostic accuracy. Modern artificial intelligence methods such as deep learning have potential supplement pathologist expertise ensure constant We developed a computational approach based on that automatically scores HER2, defines patient eligibility anti-HER2 targeted therapies breast In cohort 71 tumour resection samples, automated showed concordance 83% pathologist. The twelve discordant were then independently reviewed, leading modification diagnosis from initial assessment eight cases. Diagnostic discordance was found be largely caused perceptual differences assessing HER2 expression due high staining heterogeneity. This study provides evidence aided facilitate clinical decision making cancer identifying at risk misdiagnosis.