Japan has reportedly developed a new artificial intelligence based technology capable of detecting the prevalence of bowel cancer in minimal time. As per reports, through the currently conducted clinical experiments on the new system (test marketing), it was found that this technology could locate colorectal adenomas and benign tumors which can develop into cancer. The endoscopy performed through the innovative system confirmed that the magnified endoscopic images matched with 30,000 other images used for machine learning.
The team of researchers examined over three hundred colorectal adenomas in nearly 250 patients. Reportedly, it took lesser than a second for the system to check each magnified endoscopic image and then evaluate the tumor malignancy with 94% accuracy. Reportedly, researchers across the globe are exploring myriad AI for early cancer detection for quite a while now. As per sources, in early 2017, UK’s NHS (National Health Service) and Intel Corporation had inked a collaboration to come up with effective solutions for detecting cancer via AI. Researchers at University of Warwick came up with an online repository of the tumor and immune cells based on human tissue, for the same, while the TensorFlow architecture developed by Google is expected to form the base of this AI system and will be driven by Intel Xeon Processors.
Reports also state that research on developing a software to detect angiogenesis is being carried out by Data61 division of CSIRO (Commonwealth Scientific and Industrial Research Organization) in Australia. Scientists at Data61 and Shanghai Institute of applied physics have apparently examined 26 high resolution 3D micro-CT scans of the livers & brains of nearly twenty-six mice during different phases of cancer development and developed an algorithm that offers delivers precise geometric depictions of the blood vessels.
In the light of this scenario, cite experts, Japan’s AI-powered system is yet to establish its prominence, even though its accuracy level seems to have been exceptional.
