Gaming software inspiration for AI diagnosis system

The revelation, published in Nature Biomedical Engineering journal, originated from Sun Yat Sen University in Guangzhou China after scientists there developed an artificial neural network platform to diagnose congenital cataracts.The software, named CC-Cruiser, was developed using deep-learning algorithms, and functions on a ‘convolutional neural network’ designed to analyse how neurons are organised in the region of the brain associated with vision.{{quote-A:R-W:450-I:2-Q: The syst could potentially be used to diagnose congenital cataracts in rote areas with limited access to eye health care and professional support. -WHO:Dr Haotian Lin, Ophthalmologist and co-author of the study}}Dr Haotian Lin, an ophthalmologist and co-author of the study, said the DeepMind group, which had developed software that could beat professional human players at classic video games, had inspired th.“Since AI can play games against human players, why not create an AI that could act equally as a qualified human doctor?” she said.For the initial test, the team fed CC-Cruiser with data from 57 patients, which included 14 cases with congenital cataracts and 43 with normal eyes.The network identified cataract cases with 98.25% accuracy and also estimated the location of the cataracts with 100% accuracy.It could also suggest proper treatments with an accuracy of 92.86%.Then, another test was conducted that mirrored real-life congenital cataract scenarios – 300 photos of children with normal eyes and three with congenital cataract – which the syst passed with 100% accuracy.A website has been developed by the researchers where hospitals or eye clinics can upload data over a cloud-based server for the neural network to analyse.Lin said the syst could potentially be used to diagnose congenital cataracts in rote areas with limited access to eye health care and professional support.However, clinical trials are still needed before the AI is put into regular practice.{{image3-a:c-w:1060}}