IIT Roorkee team develops a new algorithm for brain tumours

T V Jayan New Delhi | Updated on June 11, 2020 Published on June 11, 2020

Indian computer scientists working with an NRI counterpart in Japan have developed a new machine-learning algorithm that brings better accuracy in classifying a common type of brain tumours into low or high-grade categories, arming medical oncologists with better tools to offer more effective treatment.

The new approach developed by the researchers from the Indian Institute of Technology (IIT) Roorkee led by computer science professor Raman Balasubramanian is significant because the segregation of the brain tumours – gliomas – is key to get the treatment right. Gliomas are brain tumours affecting glial cells that provide support and insulation for neurons and account for nearly a quarter of all brain tumours.

According to GLOBOCAN 2018, nearly 3 lakh new cases of brain and central nervous system cancers diagnosed worldwide with a high mortality rate of 81 per cent.

One of the fastest-growing brain tumours, gliomas are divided into four grades. While grade 1 and 2 gliomas are popularly known as low-grade gliomas, the other two are high-grade gliomas. The choice of patient treatment largely depends on being able to determine the glioma's grading. “The classifying them properly is important because each has distinctly different treatment regimen,” said Balasubramanian. Their research findings recently appeared in the journal IEEE Access.

Typically, radiologists obtain a considerable amount of data from MRI scans to reconstruct a 3D image of the scanned tissue. Much of the data available in MRI scans cannot be detected by the naked eye, such as details related to the tumour shape, texture, or the image's intensity. Artificial intelligence (AI ) algorithms are normally used to extract this data. Medical oncologists have been using this approach, called radiomics, to improve patient diagnoses, but accuracy still needs to be enhanced.

The algorithms currently in use are mostly based on AI from the 1990s or 2000s. As a result, their accuracies range from 80 to 95 per cent. There is a need for developing better algorithms that can extract more information from the medical images, said the IIT Roorkee professor whose PhD students – Rahul Kumar and Ankur Gupta – contributed mostly to work. According to him, the new algorithm could provide an accuracy of close to 98 per cent.

For the present work, the IIT researchers collaborated with Ganesh Namasivayam Pandian, a bioengineer at the Institute for Integrated Cell-Materials Sciences, Kyoto University in Japan.

The team used a dataset from MRI scans belonging to 210 people with high-grade gliomas and another 75 with low-grade gliomas. They developed an approach called CGHF, which stands for a computational decision support system for glioma classification using hybrid radiomics and stationary wavelet-based features. They chose specific algorithms for extracting features from some of the MRI scans and then trained another predictive algorithm to process this data and classify the gliomas. They then tested their model on the rest of the MRI scans to assess its accuracy.

Balasubramanian said the team is already working on a revised approach based on deep learning. This would require data from thousands of MRI images, but has the potential to improve the accuracy further.

The IIT Roorkee team which is already working with All India Institute of Medical Sciences at Rishikesh on some other medical imaging project hopes to get sufficient imaging data so that they can work on the new deep learning algorithm.

Published on June 11, 2020
This article is closed for comments.
Please Email the Editor