Generative Chromoendoscopy in Gastric Neoplasms Accurate visualization of gastric neoplasms remains critical for early diagnosis, yet conventional white-light imaging (WLI) often falls short. This multicenter prospective study introduces generative chromoendoscopy , a novel deep learning–based image transformation technique that synthetically enhances endoscopic images to improve lesion visibility. Conducted across three Japanese centers, the study included 60 patients with early gastric cancer or adenoma who underwent sequential imaging using WLI, generative chromoendoscopy, and real chromoendoscopy.
Lesion visibility was independently assessed by two endoscopists using a 7-point Likert scale. Generative chromoendoscopy demonstrated improved visibility compared to WLI and comparable performance to real chromoendoscopy, with mean visibility scores of 0.67 vs 0.51, respectively (mean…
