BreastPathQ Challenge
Neoadjuvant systemic therapy (NAT) for breast cancer (BC) is used to treat locally advanced BC prior to breast-conserving surgery. The response of cancer to NAT provides an opportunity to monitor the effectiveness of therapy, and a pathological complete response has been shown to predict survival and local recurrence. More recently a quantitative measure of response, the RCB, has been shown to provide valuable prognostic information but it is time-consuming to evaluate. The BreastPathQ challenge, which was carried out as part of the 2019 SPIE Medical Imaging Conference, addressed the problem of automatically quantifying percentage cellularity in digital breast pathology images.
Data was provided by Sunnybrook Health Sciences Centre, Toronto, Canada and the training, validation, and testing sets consisted of 2394, 185, and 1119 image patches originating from 63, 6, and 27 scanned pathology slides from 33, 4 and 18 patients, respectively. Pathologists provided ground truth labels for each patch indicating the % cellularity score. A total of 36 teams developed, validated and tested their tumor cellularity estimation algorithms during the challenge and the leading teams were able to achieve comparable accuracy to a second pathologist on the test dataset.
The challenge was supported by the International Society for Optical and Photonics (SPIE), American Association of Physicist in Medicine (AAPM), the National Cancer Institute (NCI) and the U.S. Food and Drug Administration (FDA). Data collection and curation was funded by the Canadian Cancer Society.