Artificial intelligence and digital pathology

Abstract This chapter describes recent achievements and current limitations in the field. In particular, while several AI systems show promising performance in initial testing, many have not been rigorously validated in external datasets and are not ready for clinical implementation in their current form. Currently, a lack of understanding of how deep learning systems obtain specific predictions limits the use of AI in medicine and thus revealing precise explanations will be essential to increase the use of AI in the clinic. While extensive utilization of AI systems in the clinical management of CRCs still seems years away, some systems have shown great promise, and controlled use in the clinic might be appropriate, particularly when it can be expected to improve the treatment of CRC patients.