A prospective multicenter trial reported that a fiber-optic Raman spectroscopy AI system (SPECTRA IMDx) improved real-time in vivo diagnosis of high-risk gastric neoplasia compared with routine white-light endoscopy. In the study of 198 patients and 238 lesions, the AI platform delivered higher sensitivity, specificity, and overall accuracy. Across all lesions, SPECTRA IMDx reached 90.0% sensitivity versus 58.0% for white-light endoscopy, while specificity was 81.4% versus 58.5%. Overall diagnostic accuracy increased to 83.2% from 58.4%, with statistical differences reported as significant across comparisons. In a paired subset comparing Raman AI to combined white-light and blue laser imaging, accuracy was similar, indicating the Raman tool may offer value particularly in contexts where conventional imaging is less informative. Researchers used histopathology as the reference standard during active endoscopic assessment.
Get the Daily Brief