Monday, May 20, 2024

Cracking the MS code

Students of IIT Madras have developed an algorithm that can help detect multiple sclerosis, a chronic disease affecting the nervous system and affecting 2.5 million people globally

By Murali Krishnan

Group Captain Prabhal Malakar suffers from primary-progressive multiple sclerosis (PPMS) characterized by a steady worsening of neurologic functioning. He is 59, but for the last 18 years, he has had to counter a disease which is increasing in India and in the world.
The earlier belief that MS in India was vastly different from that seen in the West has given way to the realization that they are more or less the same with minor differences.

MS has increased in India in recent years and estimates are that there are between one lakh to two lakh such patients in the 18-35 age group.The Multiple Sclerosis Foundation estimates that more than four lakh people in the US and about 2.5 million people around the world have MS. About 200 new cases are diagnosed each week in the US and radiologists have discovered that rates of MS are higher farther from the equator.

Malakar, who is also the honorary secretary of the Delhi chapter of Multiple Sclerosis Society of India, is happy to learn about the new research done by engineers for timely recognition. “The earlier you detect a debilitating disease like MS, the better for the treatment to start. The engineers have done marvelous work on this. Time is of essence for any treatment,” says Malakar.

These engineers are students of Indian Institute of Technology (IIT) Madras and have developed algorithms using artificial intelligence that will help in improving diagnosis and treatment of glioma, a malignant tumour of the glial tissue of the nervous system and MS. The algorithms for image analysis are basically a tool for diagnosis and aid clinicians in judging the progression of the disease and the usefulness of therapy.

There are four categories of MS: relapsing remitting multiple sclerosis (RRMS), secondary progressive multiple sclerosis (SPMS), primary progressive multiple sclerosis (PPMS) and progressive relapsing multiple sclerosis (PRMS). The commonest type of MS is RRMS.

Symptoms vary from patient to patient, but people with MS can suffer from fatigue or electric-shock sensations and have problems with vision, muscle control, balance and speech. At times, attacks are followed by periods of limited or complete recovery, but they can progress to permanent disability.

“These algorithms can be used for automatic analysis of MRI images in large scale clinical trials and research studies. The accuracy of these algorithms has been evaluated independently,” says Ganapathy. Krishnamurthi, professor, engineering design, IIT-Madras. “What we have developed is a technique that will develop fast evaluation of therapeutic techniques using imaging modalities, especially MRI.”

The primary research has been to develop algorithms that separate the images of tumors or lesions from the background. This would help machines identify lesions and troubled spots in the images and define them from the rest of the image.

The team has used a technology known as “Deep Learning”, which is inspired by advances in neuroscience and is loosely based on the understanding of information processing and communication within the nervous system. It is an advanced branch of study called “neural networks” that seeks to enable machines to simulate human beings in recognizing pictures and sounds through interpretations of how the brain processes them.

“Right now, our algorithm has been evaluated on a very small data set and based on that we are able to tell that this algorithm is doing well and giving us good results,” says Suthirth Vaidya, a student from the engineering team. Vaidya believes if the performance of this algorithm is put to use on a large data set, doctors will be able to use it on a day-to-day basis.

While human beings can easily identify a person in different photographs, to enable a machine to do so would need very complex algorithms. If these algorithms are in place, the accuracy in recognizing the images is expected to be much better and faster.
The next step in this endeavor would be to test extensively the effectiveness of the software and subsequently, deploy it for use by clinical collaborators. Based on the performance in a clinical setting, the engineers will try to get regulatory approval. Since training accurate models require large amounts of data, ethical committee approvals from various hospitals would be required.

The institute is in collaboration with the Thiruvanathapuram-based Sree Chitra Thirunal Hospital for Medical Sciences and Technology and is confident of seeing the product put to use in a span of two to four years.

The team felt that accurate labeling of disorder-affected regions in the brain MRI could be a difficult affair due to its “complex shape and vague boundaries”. It is a mind-numbing task since radiologists cannot visualize in 3D and the task needs to be performed slice by slice.

These methods, when implemented, can substantially reduce the time and cost for diagnosis of various brain diseases like MS. The algorithms for image analysis are basically a tool for diagnosis and aid clinicians to judge the progression of the disease and efficacy of therapy.

Labeling of the disorder-affected regions in the brain MRI is a mind-numbing task since radiologists cannot visualise in 3-D and the task is to be performed slice by slice.

Vaidya’s colleague Abhijith Chunduru was also part of the team that won the Longitudinal Multiple Sclerosis Segmen-tation challenge at the International Symposium on Biomedical Imaging in New York a few months earlier. Chunduru believes it would indeed be possible to bring down the cost of diagnosis of MS. “This would help radiologists make very quick analysis and get quantitative numbers on how quickly or slowly the disease is progressing within patients,” says Chunduru.

MS is the most widespread disabling neurological condition of young adults around the world.

There are relapsing and remitting types of MS and progressive types, but the course is rarely predictable. Unfortunately, researchers still don’t fully understand the causes of MS or why the rate of progression is so difficult to determine despite millions of dollars being pumped into research for many years.
Krishnamurthi believes that the clinical trials to assess the effectiveness of the software is crucial.

“Yes, these algorithms will play a very good role, especially five years from now when we would have got over the bumps and problems. That is our belief,” he adds. That would come as a big relief to many MS patients.


News Update