NuCLS: A scalable crowdsourcing approach and dataset for nucleus classification and segmentation in breast cancer.

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.

Outcome disparities in colorectal cancer: a SEER-based comparative analysis of racial subgroups.

PURPOSE: Previous studies of ethnic disparities in colorectal cancer (CRC) have focused mainly on patients of Caucasian and African-American descent. We aimed to evaluate outcomes for a range of races, representing a broader demographic of the US population. METHODS: The Surveillance, Epidemiology, and End Results database was queried to identify patients with CRC diagnosed between 1994 and 2014. We performed unadjusted Kaplan-Meier test and multivariable covariate-adjusted Cox models to calculate the overall and CRC-specific survival of patients according to their race. RESULTS: We identified 401,723 patients diagnosed with CRC between 1994 and 2014. Overall survival (OS) and CRC-specific survival were compared across different races stratified by age, sex, marital status, disease stage and grade, and undergoing surgery as a treatment. Overall, Asian/Pacific Islanders and Hispanics had improved CRC-specific survival compared to Whites (HR = 0.873, 95%CI 0.853-0.893, P < .001, and HR = 0.958, 95%CI 0.937-0.979, P < .001, respectively). Blacks had the worst CRC-specific survival outcomes when compared to Whites (HR = 1.215, 95%CI 1.192-1.238, P < .001). Racial disparity persisted when looking at two different time periods (1994-2003 and 2004-2014). CONCLUSIONS: Asians/Pacific Islanders have improved outcomes from CRC compared to other races. Multifactorial, including genetic, environmental, and socioeconomic factors appear to influence outcomes and need to be addressed separately in order to reduce racial disparities among patients with CRC.