Automated Gleason Grading Challenge 2022

Prostate cancer is characterized by an abnormal growth of cells in the prostate glands. It is also the second most common cancer among men worldwide which kills 1 in 40 men annually. The progression of it is determined according to the Gleason grading system, which also serves as a guide to decide the appropriate treatment a patient should receive.

In diagnosis, tissue samples are first obtained during prostate biopsies and examined via visual inspection by  pathologists. Unfortunately, manual examination can be prone to inter-observer variability even between expert pathologists. This may result in missing a cancer diagnosis or unnecessary treatment due to over-grading. Furthermore, such methods are also time consuming as differentiation between malignant and benign biopsy samples are often in extensive amounts.

Therefore, this competition aims to utilize the potential of automated deep learning systems in the diagnosis of prostate cancer, with the ultimate goal of improving prognosis and quality of life of patients. Research studies have shown support for such artificial intelligence methods in achieving pathologist-level performance, as well as improvements in speed, accuracy, and consistency of the results. This competition will also seek to define an assessment standard for machine learning-based algorithms for reported results, as well as choosing the most effective solution for future clinical trials.

In this challenge, we publish a H&E-stained whole slide image dataset of prostatectomy and biopsy specimens with pixel-level annotations performed by experienced pathologists and Gleason Score. Additionally, we also provide a set of images scanned by multiple scanners to assess the algorithm performance of handling variations caused by image digitalization. The submitted algorithm should be accurate to detect different Gleason Patterns and also generalized to process images scanned by different scanners. To the best of our knowledge, this is the first challenge in the field of digital pathology that investigate the variations caused by image scanning.