The manual identification of gross errors is inefficient, error-prone and highly subjective.

The curation of pathologist-annotated, high-quality slide datasets is a serious bottleneck in digital pathology. These data fuel downstream analysis and the development of reliable AI-based solutions for various applications. By streamlining the production of high-quality histopathology images via fully automating tissue processing and digitization workflows it expedites data interpretation and usage and thereby accelerating preclinical toxicology studies. Our solution hosts slides on our cloud-based platform, PathologyMapTM, providing researchers with expedited viewing, storage, annotation, and a seamless connection to image analysis and pathology experts for objective analysis. The online platform has also amassed a large collection of preclinical pathology data which we’ve leveraged to develop AI-driven tools such as slide tagging, content-based image retrieval, and automated QC. This talk will focus on our development and integration of an Auto-QC system into the HistoWiz PathologyMapTM platform that identifies common quality issues observed in digitized slides such as blurriness, folds, tissue tearing, dirty slides, and air bubbles. Identification of such gross errors is manual, subjective, and time consuming. To address this, our scalable and reproducible AI-model for slide QC reduces lab workload and helps ensure consistent high-quality digitized slides for our customers.