Deep Learning and Automatic Polyp Detection in Colonoscopies: a Review of Recent Contributions and Future Outlook

Colorectal (CRC) cancer is one of the most threatening death cause, therefore its early detection and possible treatment challenges the worldwide scientists. Human accuracy in polyps, adenomas and diverticula detection depends on many subjective factors, from colon cleansing quality to the device technical conception and physician experience or degree of fatigue and attention. Automatic polyp detection has preoccupied many groups of researchers in the last decade. An overview on these last researches is provided. More recently, deep learning begun to be applied in many fields and we will try to highlight the related applications for colonoscopy, in this paper.