Scientists of Institute of Advanced Study in Science and Technology (IASST), Guwahati, have developed an artificial intelligence (AI) enabled diagnostic system that does rapid testing of oral squamous cell carcinoma as well as grading cancer. 

The algorithm for this cancer’s detection was made by the department of Central Computational and Numerical Sciences at IASST under the leadership of Dr Lipi B Mahanta. Nearly 100,000 cases of oral squamous cell carcinoma are reported every year in India. The team decided to make this indigenous testing kit when they realised that such tests didn’t exist in the country. 

The algorithm used transfer learning using pre-trained deep convolutional neural network (CNN). The team deployed four pre-trained AI models - Alexnet, VGG-16, VGG-19, and Resnet-50 - to figure out which model would be ideal for classification problem along with a proposed CNN model that would best suit the problem statements. 

While Resnet-50 had an accuracy rate of 92.15 per cent, the proposed CNN model performed better at 97.5 per cent, as compared to the transfer learning approaches deployed earlier. Currently, the group is adapting the algorithm into a software so that field trials can be conducted.

 The team’s groundbreaking work has been published in Neural Networks, the official journal of the world’s three oldest neural modelling societies: the International Neural Network Society (INNS), the European Neural Network Society (ENNS), and the Japanese Neural Network Society (JNNS).