In a path-breaking innovation that can revolutionise cancer treatment, a team of Indian researchers have developed an AI-based algorithm that can do more than just detect the disease.
Researchers from the Indian Institute of Science Education and Research,Kolkata (IISER-K) and Indian Institute of Technology, Kanpur (IIT-K) have developed an algorithm based on the light scattering properties of healthy and precancerous cells.
The structural difference between healthy and precancerous cells causes light to scatter differently. This is due to the difference in the refractive index of these two types of cells. However, these differences in the pattern of scattered light are subtle are cannot be observed by the naked eye.
This is where the new AI-based algorithm comes into play. The algorithm can not only espy the difference between healthy and precancerous tissue, but can also identify the different progressive stages of the disease with an accuracy of over 95 percent in a matter of minutes.
“The microstructure of normal tissue is uniform but as disease progresses the tissue microstructure becomes complex and different. Based on this correlation, we created a novel light scattering-based method to identify these unique microstructures for detecting cancer progression,” Sabyasachi Mukhopadhyay from IISER-K, told a newspaper while talking about this pioneering development.
In order to quantify the difference in the light scattered by the cells as the disease progresses, the research team used a statistical biomarker called Multifractal Detrended Fluctuation Analysis (MFDFA). MFDFA has two parameters – Hurst exponent and width of singularity spectrum. For distinguishing the data and classifying healthy and different grades of cancer tissues, AI-based algorithms such as hidden Markov model (HMM) and support vector machine (SVM) were employed.
Professor Prasanta K Panigrahi from IISER-K, and the co-author of papers published in collaboration with Mukhopadhyay in Journal of Biomedical Optics added, “The classification of healthy and precancerous cells becomes robust by converting the information obtained from the scattered light into characteristic tissue-specific signature. The signature captures the variations in tissue morphology.”
Trials from in vitro samples yielded results with over 95 percent accuracy while a few in vivo samples yielded over 90 percent accuracy.
“In the case of in vitro samples we were able to discriminate between grade 1 and grade 2 cancer,” said Professor Nirmalya Ghosh from IISER-K who is another collaborator of Mukhopadhyay and Panigraha.
The research team intends to expand investigations to study precancer detection using in vivo samples.
Try deep learning using MATLAB