Ehab Hafiz

BACKGROUND: Deep learning enables accurate high-resolution mapping of cells and tissue structures that can serve as the foundation of interpretable machine-learning models for computational pathology. However, generating adequate labels for these structures is a critical barrier, given the time and effort required from pathologists. RESULTS: This article describes a novel collaborative framework for engaging crowds of medical students and pathologists to produce quality labels for cell nuclei. We used this approach to produce the NuCLS dataset, containing >220,000 annotations of cell nuclei in breast cancers. This builds on prior work labeling tissue regions to produce an integrated tissue region- and cell-level annotation dataset for training that is the largest such resource for multi-scale analysis of breast cancer histology. This article presents data and analysis results for single and multi-rater annotations from both non-experts and pathologists. We present a novel workflow that uses algorithmic suggestions to collect accurate segmentation data without the need for laborious manual tracing of nuclei. Our results indicate that even noisy algorithmic suggestions do not adversely affect pathologist accuracy and can help non-experts improve annotation quality. We also present a new approach for inferring truth from multiple raters and show that non-experts can produce accurate annotations for visually distinctive classes. CONCLUSIONS: This study is the most extensive systematic exploration of the large-scale use of wisdom-of-the-crowd approaches to generate data for computational pathology applications.

The limited preservation duration of organs has contributed to the shortage of organs for transplantation. Recently, a tripling of the storage duration was achieved with supercooling, which relies on temperatures between -4 and -6 °C. However, to achieve deeper metabolic stasis, lower temperatures are required. Inspired by freeze-tolerant animals, we entered high-subzero temperatures (-10 to -15 °C) using ice nucleators to control ice and cryoprotective agents (CPAs) to maintain an unfrozen liquid fraction. We present this approach, termed partial freezing, by testing gradual (un)loading and different CPAs, holding temperatures, and storage durations. Results indicate that propylene glycol outperforms glycerol and injury is largely influenced by storage temperatures. Subsequently, we demonstrate that machine perfusion enhancements improve the recovery of livers after freezing. Ultimately, livers that were partially frozen for 5-fold longer showed favorable outcomes as compared to viable controls, although frozen livers had lower cumulative bile and higher liver enzymes.