A Pathologist-Annotated Dataset for Validating Artificial Intelligence: A Project Description and Pilot Study

Dudgeon, Sarah N; Wen, Si; Hanna, Matthew G; Gupta, Rajarsi; Amgad, Mohamed; Sheth, Manasi; Marble, Hetal; Huang, Richard; Herrmann, Markus D; Szu, Clifford H.; Tong, Darick; Werness, Bruce; Szu, Evan; Larsimont, Denis; Madabhushi, Anant; Hytopoulos, Evangelos; Chen, Weijie; Singh, Rajendra; Hart, Steven N.; Saltz, Joel; Salgado, Roberto; Food, Brandon D Gallas United States; Administration, Drug; Devices, Center for; Health, Radiological; Science, Office of; Laboratories, Engineering; Reliability, Division of Imaging DiagnosticsSoftware; Oak, White; MD,; Center, Memorial Sloan Kettering Cancer; York, New; NY,; Informatics, Stony Brook Medicine Dept of Biomedical; Brook, Stony; Pathology, Department of; University, Northwestern; Building, Rubloff; Illinois, Chicago; Food, United States; Quality, Office of Product; Evaluation,; Evidence, Office of Clinical; Analysis,; Biostatistics, Division of; School, Massachusetts General HospitalHarvard Medical; Boston,; MA,; Pathology, Computational; Hospital, Massachusetts General; School, Harvard Medical; Origin, Arrive; Francisco, San; CA,; Bordet, Institut Jules; Brussels,; Belgium,; University, Case Western Reserve; Cleveland,; OH,; Inc., iRhythm Technologies; health, Northwell; Medicine, Zucker School of; Research, Department of Health Sciences; Clinic, Mayo; Rochester, MN; Research, Division of; Centre, Peter Mac Callum Cancer; Melbourne,; Australia,; Hospitals, GZA-ZNA; Antwerp,

Purpose: In this work, we present a collaboration to create a validation dataset of pathologist annotations for algorithms that process whole slide images (WSIs). We focus on data collection and evaluation of algorithm performance in the context of estimating the density of stromal tumor infiltrating lymphocytes (sTILs) in breast cancer. Methods: We digitized 64 glass slides of hematoxylin- and eosin-stained ductal carcinoma core biopsies prepared at a single clinical site. We created training materials and workflows to crowdsource pathologist image annotations on two modes: an optical microscope and two digital platforms. The workflows collect the ROI type, a decision on whether the ROI is appropriate for estimating the density of sTILs, and if appropriate, the sTIL density value for that ROI. Results: The pilot study yielded an abundant number of cases with nominal sTIL infiltration. Furthermore, we found that the sTIL densities are correlated within a case, and there is notable pathologist variability. Consequently, we outline plans to improve our ROI and case sampling methods. We also outline statistical methods to account for ROI correlations within a case and pathologist variability when validating an algorithm. Conclusion: We have built workflows for efficient data collection and tested them in a pilot study. As we prepare for pivotal studies, we will consider what it will take for the dataset to be fit for a regulatory purpose: study size, patient population, and pathologist training and qualifications. To this end, we will elicit feedback from the FDA via the Medical Device Development Tool program and from the broader digital pathology and AI community. Ultimately, we intend to share the dataset, statistical methods, and lessons learned.