Using HistoQC to predict interobserver disagreement on HER2 assessment

The study investigates the impact of digital image quality in Anatomical Pathology on diagnostic agreement among pathologists, focusing on HER2 diagnosis in cases of invasive breast carcinoma. A total of 132 slides were digitized using the Aperio CS2 scanner and the diagnoses were independently made by two pathologists in a double-blind fashion, with a third reader intervening in cases of disagreement.
Open-source HistoQC software was employed to extract quality indicators from the digital images. These quality indicators were then employed as inputs for a classifier designed to identify cases where pathologists disagreed on the HER2 diagnosis, indicating instances of challenging interpretation.
The results of the study demonstrated a noteworthy correlation between parameters reflecting digital image quality and inter-observer disagreements. This suggests that certain aspects of digital slide quality play a substantial role in influencing discrepancies in interpretation among pathologists, especially in the context of HER2 diagnosis in breast cancer cases. The findings underscore the importance of considering and optimizing digital image quality for enhancing diagnostic agreement in Anatomical pathology.