Gastric cancer in China can be characterized with high incidence rate, low early diagnosis rate and high mortality rate. Its prognosis is closely tied to the timing of diagnosis. In patients with early stage gastric cancer, the post-surgery 5-year survival rate is over 90%.
In general, tens of thousands of endoscopic images are captured each time the capsule endoscopy is performed. Common gastric lesions that can be identified in endoscopic images include:
As an early screening tool, the interpretation of massive endoscopic images has become a major challenge for clinicians. How to quickly and accurately identify lesions is key to the success of AI enabled endoscopic image grading. In addition to achieving high sensitivity and specificity for the deep learning algorithms, it is imperative to develop an effective quality management system to minimize risks by way of a centralized reading center.