Long overdue national big data policies hinder accurate and equitable cancer detection AI systems

National, compulsory cancer registry organizations (e.g. Canada CCR, Mexico NCR, U.K. NCRAS, U.S.A. SEER) collect tabular but not imaging cancer data that are critical to optimize the accuracy and consistency of the increasingly-pervasive artificial intelligence (AI) medical systems that are approved by regulatory agencies (e.g. Health Canada and the U.S.A. Food and Drug Administration), but typically with insufficient image test data [1]. Moreover, commercial hospital-tested computer-aided diagnosis (CAD) screen for malignant lesions from medical images [2], but have not advanced further to other algorithms to determine the cancer stage or recommend best treatment because there are insufficient high-quality data collections of both imaging data and their corresponding tabular medical data.
Source: Journal of Medical Imaging and Radiation Sciences - Category: Radiology Authors: Tags: Commentary Source Type: research