New artificial intelligence system can quickly detect colorectal cancer

Scientists have developed an artificial intelligence (AI) system that can automatically detect colorectal cancer in less than one second. Researchers at Showa University in Yokohama, Japan, developed a computer-aided diagnostic system that uses an image (a 500-fold magnification view of a colorectal polyp) and analyzes about 300 polyp features after applying a narrow-band imaging (NBI) model.

The system compares the characteristics of each polyp with more than 30,000 images for machine learning, allowing it to predict lesion pathology in less than one second.
Researchers studied 250 men and women, of whom colorectal polyps were detected. The AI ​​helper system is used to predict the pathology of each polyp and compare these predictions with the pathology report obtained from the final resected specimen.
The team used real-time evaluation of meat using an AI-assisted system, providing sensitivity of 94%, specificity of 79%, accuracy of 86%, positive and negative predictive values ​​of 79% and 93%, respectively, to identify tumors.
Yokohama Showa University stated: "We believe these results are acceptable for clinical applications. Our goal is to obtain regulatory approval for diagnostic systems."