A Systematic Collection of Medical Image Datasets for Deep Learning

• DOI: 10.48550/arxiv.2106.12864
• Publication Date: 2021-01-01
• Authors: Johann Li; Guangming Zhu; Cong Hua; Mingtao Feng; Basheer Bennamoun; Ping Li; Xiaoyuan Lu; Juan Song; Peiyi Shen; Xu Xu; Lin Mei; Liang Zhang; Syed Afaq Ali Shah; Mohammed Bennamoun
• Keywords: Robustness; Medical Research

The astounding success made by artificial intelligence (AI) in healthcare and other fields proves that AI can achieve human-like performance. However, always comes with challenges. Deep learning algorithms are data-dependent require large datasets for training. lack of data the medical imaging field creates a bottleneck application deep to image analysis. Medical acquisition, annotation, analysis costly, their usage is constrained ethical restrictions. They also many resources, such as human expertise funding. That makes it difficult non-medical researchers have access useful data. Thus, comprehensive possible, this paper provides collection associated challenges research. We collected information around three hundred mainly reported between 2013 2020 categorized them into four categories: head & neck, chest abdomen, pathology blood, ``others''. Our has purposes: 1) provide most up date complete list be used universal reference easily find clinical analysis, 2) guide on methodology test evaluate methods' performance robustness relevant datasets, 3) ``route'' topics, challenge leaderboards.