Biophotonics and artificial intelligence in diagnostic endoscopy of the digestive system
Biophotonics and artificial intelligence in diagnostic endoscopy of the digestive system
Endoscopic imaging of the mucosa constitutes the diagnostic cornerstone in the management of gastrointestinal diseases. In the oncological context, endoscopy is primarily used to screen the mucosa for lesions and to determine their nature as benign or malignant.. Beyond its role in lesion characterization, endoscopy also plays a key role in cancer surveillance. Patients at increased risk of gastrointestinal malignancies are enrolled in surveillance programs with defined intervals to facilitate early detection of carcinomas or their precursor lesions, such as adenomatous or sessile serrated polyps (AP/SSP) in the colon and low- or high‑grade dysplasia (LGD/HGD) in Barrett’s esophagus. However, diminutive and flat adenomas are easily overlooked in the colon, and within the Barrett’s segment, distinguishing low‑ or high‑grade dysplasia from non‑dysplastic mucosa can be difficult. In the biliary tree, biliary strictures are also difficult to characterize.
This thesis of Jouke van der Laan aims to address the challenges in detecting neoplastic lesions the esophagus, colon, and biliary tract by exploring innovations in endoscopic imaging to enhance tissue assessment. Therefore, advanced optical technologies based on biophotonics, such as quantitative fluorescence molecular endoscopy and light-scattering based measurements, are clinically investigated to extract additional molecular information during procedures. Particularly, we investigated the concept of field cancerization to predict the presence of cancer based on scattering measurements of non-dysplastic tissue. Employing this phenomenon could improve risk stratification during surveillance patients with Barrett’s esophagus. In parallel, artificial intelligence is explored for facilitating the implementation of endoscopic innovations and for augmenting human interpretation of endoscopic data.