A pathologist-AI collaboration framework for enhancing diagnostic accuracies and efficiencies.

In pathology, the deployment of artificial intelligence (AI) in clinical settings is constrained by limitations in data collection and in model transparency and interpretability. Here we describe a digital pathology framework, nuclei.io, that incorporates active learning and human-in-the-loop real-time feedback for the rapid creation of diverse datasets and models. We validate the effectiveness of the framework via two crossover user studies that leveraged collaboration between the AI and the pathologist, including the identification of plasma cells in endometrial biopsies and the detection of colorectal cancer metastasis in lymph nodes. In both studies, nuclei.io yielded considerable diagnostic performance improvements. Collaboration between clinicians and AI will aid digital pathology by enhancing accuracies and efficiencies.

Does Milan affect management? A retrospective analysis of resection rate and time to surgery among Milan categories.

INTRODUCTION: The Milan System for Reporting Salivary Gland Cytopathology (MSRSGC) is an established system with reproducible risk of malignancies (ROM) for salivary gland fine needle aspiration (SGFNA). No studies have reviewed the relationship between Milan categories and the resection rate (RR) and time to resection (TTR). METHODS: We searched our database (January 1, 2011 to January 4, 2021) for non-lymphoma SGFNAs and assigned appropriate MSRSGC categories. RR and TTR were calculated and compared for each category. A literature search was performed; RRs and TTRs were compared. RESULTS: Seven hundred and eighty SGFNAs were identified, 333 with follow-up. RR was highest in suspicious for malignancy (SUS, V; 70.6%, n = 12/17), followed by the salivary gland neoplasm of uncertain malignant potential (SUMP, IVb; 69.6%, n = 80/115) and malignant (M, VI; 55.6%, n = 75/135). Among M, primary tumors had a higher RR (65.1%, n = 41/63) than metastases (47.2%, n = 34/72, p = .36). In literature review, SUS had the highest RR (69.3%, n = 233/336) followed by M (61.6%, n = 821/1332) and SUMP (60.2%, n = 632/1050). TTR was shorter in SUS (mean = 32.3 days, median = 25 days). Within the benign neoplasms (BN, IVa), Pleomorphic adenomas (PAs) had a higher RR than Warthin tumors (WTs) (66.3% vs. 37.2%, p < .00001), and a shorter TTR (median = 63 days vs. 90 days). CONCLUSIONS: Tumors classified as SUS had higher RR and at shorter intervals than those classified as SUMP. PAs have higher RRs and more expedient surgery than WTs. Cases classified as M are less likely to undergo follow-up than SUS, perhaps due to a lower RR for metastases.