Small Bowel Capsule Endoscopy and artificial intelligence: First or second reader?

• Published in: Baillière's best practice & research. Clinical gastroenterology Vol. 52-53; p. 101742
• Main Authors: Piccirelli, Stefania, Milluzzo, Sebastian Manuel, Bizzotto, Alessandra, Cesaro, Paola, Pecere, Silvia, Spada, Cristiano
• Format: Journal Article
• Language: English
• Published: Kidlington Elsevier Ltd 01.06.2021; Elsevier Limited
• Subjects: Accuracy; Agreements; Algorithms; Artificial intelligence; Automatic reading; Capsule endoscopy; CNN-based algorithm; Deep learning; Endoscopy; Neural networks; Patients; Polyps; Reading; Small intestine; Systematic review; Tumors; Ulcers; Wavelet transforms; Automatic reading; Deep-learning; Capsule endoscopy; Artificial intelligence; CNN-based algorithm
• ISSN: 1521-6918; 1532-1916; 1532-1916
• DOI: 10.1016/j.bpg.2021.101742

Several machine learning algorithms have been developed in the past years with the aim to improve SBCE (Small Bowel Capsule Endoscopy) feasibility ensuring at the same time a high diagnostic accuracy. If past algorithms were affected by low performances and unsatisfactory accuracy, deep learning systems raised up the expectancy of effective AI (Artificial Intelligence) application in SBCE reading. Automatic detection and characterization of lesions, such as angioectasias, erosions and ulcers, would significantly shorten reading time other than improve reader attention during SBCE review in routine activity. It is debated whether AI can be used as first or second reader. This issue should be further investigated measuring accuracy and cost-effectiveness of AI systems. Currently, AI has been mostly evaluated as first reader. However, second reading may play an important role in SBCE training as well as for better characterizing lesions for which the first reader was uncertain. Capsule Endoscopy (CE) is a well-established non-invasive diagnostic method able to provide a reliable visualization and assessment of small bowel diseases. In the last years, Artificial Intelligence (AI) spread out in clinical imaging, involving also the field of endoscopy. Small Bowel Capsule Endoscopy (SBCE) might be considered as a suitable field to be computer-assisted, since it is burdened by a long reading time, being fatigue a potential reason of missing relevant findings. In this field, AI could be used as a first-time reader, functioning as a preliminary filter able to reduce reading time. Otherwise, AI could serve as a second-time reader, confirming or denying findings detected by the endoscopist.