According to a study published by members at the Radiology and Pathology & Laboratory Medicine in the Perelman School of Medicine at the University of Pennsylvania, an emerging technique called federated learning lauded to be a solution to how machine learning can be trained on diverse datasets without sharing critical patient data that can cause legal, privacy and cultural challenges.

Federated learning was first implemented by Google for the keyboard's autocorrect functionality. It enabled the algorithm to be trained on local data samples on multiple decentralised devices or servers, without exchanging them. For the first time, this technique is being used for enabling medical professionals to exchange data across institutes, facilitating Artificial Intelligence systems have better answers to medical questions.

The researchers at Penn Medicine researchers have piloted the use of this technique in the context of brain imaging, by analysing MRI scans of patients with a brain tumour and distinguishing the cancerous region from healthy brain tissues.

The model can be distributed to various hospitals across the world where other doctors can share their AI model on top on this shared model by adding their own patients' MRI scans. This new model will then be included in a centralised server, eventually to be restored within a unified model that has drawn from knowledge across hospitals worldwide.

"The more data the computational model sees, the better it learns the problem, and the better it can address the question that it was designed to answer," Bakas said. "Traditionally, machine learning has used data from a single institution, and then it became apparent that those models do not perform or generalise well on data from other institutions."

The model currently is being tested by the U.S. Food and Drug Administration, where it will undergo validation and approval systems. Post these steps; the model will have a commercial license, enabling hospitals to use it as a clinical tool. The tool will especially be helpful to medical professionals such as radiologists, radiation oncologists, and neurosurgeons make important decisions about patient care.

In India, brain tumour is ranked as the 10th most common kind of tumour affecting Indians. There is a high death rate in India - yearly there are 28,000 cases of brain tumours reported in India, out of which more than 24,000 people reportedly die