Reproducible Reporting of the Collection and Evaluation of Annotations for Artificial Intelligence Models.

This work puts forth and demonstrates the utility of a reporting framework for collecting and evaluating annotations of medical images used for training and testing artificial intelligence (AI) models in assisting detection and diagnosis. AI has unique reporting requirements, as shown by the AI extensions to the Consolidated Standards of Reporting Trials (CONSORT) and Standard Protocol Items: Recommendations for Interventional Trials (SPIRIT) checklists and the proposed AI extensions to the Standards for Reporting Diagnostic Accuracy (STARD) and Transparent Reporting of a Multivariable Prediction model for Individual Prognosis or Diagnosis (TRIPOD) checklists. AI for detection and/or diagnostic image analysis requires complete, reproducible, and transparent reporting of the annotations and metadata used in training and testing data sets. In an earlier work by other researchers, an annotation workflow and quality checklist for computational pathology annotations were proposed. In this manuscript, we operationalize this workflow into an evaluable quality checklist that applies to any reader-interpreted medical images, and we demonstrate its use for an annotation effort in digital pathology. We refer to this quality framework as the Collection and Evaluation of Annotations for Reproducible Reporting of Artificial Intelligence (CLEARR-AI).

Training pathologists to assess stromal tumour-infiltrating lymphocytes in breast cancer synergises efforts in clinical care and scientific research.

A growing body of research supports stromal tumour-infiltrating lymphocyte (TIL) density in breast cancer to be a robust prognostic and predicive biomarker. The gold standard for stromal TIL density quantitation in breast cancer is pathologist visual assessment using haematoxylin and eosin-stained slides. Artificial intelligence/machine-learning algorithms are in development to automate the stromal TIL scoring process, and must be validated against a reference standard such as pathologist visual assessment. Visual TIL assessment may suffer from significant interobserver variability. To improve interobserver agreement, regulatory science experts at the US Food and Drug Administration partnered with academic pathologists internationally to create a freely available online continuing medical education (CME) course to train pathologists in assessing breast cancer stromal TILs using an interactive format with expert commentary. Here we describe and provide a user guide to this CME course, whose content was designed to improve pathologist accuracy in scoring breast cancer TILs. We also suggest subsequent steps to translate knowledge into clinical practice with proficiency testing.

Slide-to-Slide Tissue Transfer and Array Assembly From Limited Samples for Comprehensive Molecular Profiling.

Tissue microarrays (TMA) have become an important tool in high-throughput molecular profiling of tissue samples in the translational research setting. Unfortunately, high-throughput profiling in small biopsy specimens or rare tumor samples (eg, orphan diseases or unusual tumors) is often precluded owing to limited amounts of tissue. To overcome these challenges, we devised a method that allows tissue transfer and construction of TMAs from individual 2- to 5-µm sections for subsequent molecular profiling. We named the technique slide-to-slide (STS) transfer, and it requires a series of chemical exposures (so-called xylene-methacrylate exchange) in combination with rehydrated lifting, microdissection of donor tissues into multiple small tissue fragments (methacrylate-tissue tiles), and subsequent remounting on separate recipient slides (STS array slide). We developed the STS technique by assessing the efficacy and analytical performance using the following key metrics: (a) dropout rate, (b) transfer efficacy, (c) success rates using different antigen-retrieval methods, (d) success rates of immunohistochemical stains, (e) fluorescent in situ hybridization success rates, and (f) DNA and (g) RNA extraction yields from single slides, which all functioned appropriately. The dropout rate ranged from 0.7% to 6.2%; however, we applied the same STS technique successfully to fill these dropouts ("rescue" transfer). Hematoxylin and eosin assessment of donor slides confirmed a transfer efficacy of >93%, depending on the size of the tissue (range, 76%-100%). Fluorescent in situ hybridization success rates and nucleic acid yields were comparable with those of traditional workflows. In this study, we present a quick, reliable, and cost-effective method that offers the key advantages of TMAs and other molecular techniques-even when tissue is sparse. The perspectives of this technology in biomedical sciences and clinical practice are promising, given that it allows laboratories to create more data with less tissue.