Artificial Intelligence-Based Polyp Detection in Colonoscopy: Where Have We Been, Where Do We Stand, and Where Are We Headed?

Abstract

<b><i>Background:</i></b> In the past, image-based computer-assisted diagnosis and detection systems have been driven mainly from the field of radiology, and more specifically mammography. Nevertheless, with the availability of large image data collections (known as the “Big Data” phenomenon) in correlation with developments from the domain of artificial intelligence (AI) and particularly so-called deep convolutional neural networks, computer-assisted detection of adenomas and polyps in real-time during screening colonoscopy has become feasible. <b><i>Summary:</i></b> With respect to these developments, the scope of this contribution is to provide a brief overview about the evolution of AI-based detection of adenomas and polyps during colonoscopy of the past 35 years, starting with the age of “handcrafted geometrical features” together with simple classification schemes, over the development and use of “texture-based features” and machine learning approaches, and ending with current developments in the field of deep learning using convolutional neural networks. In parallel, the need and necessity of large-scale clinical data will be discussed in order to develop such methods, up to commercially available AI products for automated detection of polyps (adenoma and benign neoplastic lesions). Finally, a short view into the future is made regarding further possibilities of AI methods within colonoscopy. <b><i>Key Messages:</i></b> Research<b><i></i></b>of<b><i></i></b>image-based lesion detection in colonoscopy data has a 35-year-old history. Milestones such as the Paris nomenclature, texture features, big data, and deep learning were essential for the development and availability of commercial AI-based systems for polyp detection.

Artificial Intelligence in Medicine: Passing Hype or the Holy Grail of Solutions?

“…These examples show how patients’ safety could be remarkably improved in the future, but also demonstrate the need for the broad and open-minded provision of clinical data for subsequent analysis. The more data available, the better the trained algorithm will be and can thus, in turn, truly support our clinical work, as has been delineated for computer-based lesion detection in colonoscopy by Wittenberg and Rai­thel [13]. AI-augmented methods will prove themselves to be a selection criterion, and those who offer them will outperform others.…”

“…Standardization does not necessarily mean the uniform treatment of patients, but rather the adherence to general principles and the use of established classifications, medical ontologies, and concerted processes. The commonly used classification systems like the APACHE score for sepsis [14], the CHICAGO classification for esophageal motility [8], and the PARIS classification for gastrointestinal adenomas [13] all served as primers for the establishment of artificial neural networks in these fields. Likewise, the standardization of surgical processes, even at a very low level as proposed for cholecystectomy by Immenroth et al [15], is necessary for the computer-based interpretation of operative interventions and workflow recognition.…”