To validate different methods for computing TC, representative sections from 64 patients with residual invasive BC on resection specimens following NAT were acquired. Representative routine Hematoxylin and Eosin glass slides were scanned at 20X magnification (0.5 lm/pixel) using an Aperio scanner. The study was approved by the institutional ethics board and an amendment to share the data was obtained. Clinical data corresponding to each patient includes: Age at diagnosis Menopausal status NAT type (chemo/HER2/RAD/Endocrine) Histological type ER/PR/HER2 status LN status A breast pathology fellow annotated identified patches on a digital pathology viewing platform, Sedeen Viewer . For each patch, a TC score, ranging from 0% to 100% for assessment of RCB , was provided. Patches which did not contain any tumour cells were assigned a TC score of 0%. Annotations for >30,000 nuclei (3,868 lymphocyte, 10,407 benign epithelial, and 16,419 malignant epithelial figures) were also marked by the pathologist from 166 ROIs in sets A and B which included a mixture of lymphocyte, benign, and malignant epithelial nuclei figures. This data was used for the BreastPathQ challenge co-sponsored by SPIE/AAPM/NCI and currently 338 participants have registered on the CodaLab site to download patches from these WSIs. The aim of the data collection ws to train and validate methods of assessing residual tumour cellularity after NAT. We also have annotations for malignant and benign epithelial cells and TILs which are valuable for assessing performance of cell classification methods. This collection is distinct from other collections as it is focused on post-NAT patients with residual tumour and was collected to allow automatic methods of residual cancer burden to be developed. There are many histopathological changes that occur after therapy and the image data is linked with information about receptor status and treatment type making this a unique resource. |