Subjective and objective quality assessment of gastrointestinal endoscopy images: From manual operation to artificial intelligence

• DOI: 10.3389/fnins.2022.1118087
• Publication Date: 2023-02-14T17:36:48Z
• Authors: Peng Yuan; Ruxue Bai; Yan Yan; Shijie Li; Jing Wang; Changqi Cao; Qi Wu
• Keywords: Motion blur

Gastrointestinal endoscopy has been identified as an important tool for cancer diagnosis and therapy, particularly treating patients with early gastric (EGC). It is well known that the quality of gastroscope images a prerequisite achieving high detection rate gastrointestinal lesions. Owing to manual operation detection, in practice, it possibly introduces motion blur produces low-quality during imaging process. Hence, assessment key process endoscopy. In this study, we first present novel image (GIMB) database includes 1,050 generated by imposing 15 distortion levels on 70 lossless associated subjective scores produced viewers. Then, design new artificial intelligence (AI)-based evaluator (GIQE) leverages newly proposed semi-full combination subspace learn multiple kinds human visual system (HVS) inspired features providing objective scores. The results experiments conducted GIMB confirm GIQE showed more effective performance compared its state-of-the-art peers.