Short Bytes: The researchers at the MRC LMS have worked out a new machine learning software that can transform the field of cardiology. The software analyses MRI scan reports and creates 3D virtual images of the heart. The software can predict heart failure rates of a patient with an accuracy of 80 percent.According to a breakthrough research published in the Radiology journal, the team at the MRC London Institute of Medical Sciences have trained a machine learning software. This can analyze blood tests and heart beats to predict the possible failure of the human heart.
The MRI scans and blood test results of around 250 patients – diagnosed with pulmonary hypertension – were fed to the software during the training phase. The software takes a look at the heart images captured during the MRI scan and uses advanced image processing techniques to create a virtual 3D heart. It focuses on 30,000 different points in the heart’s structure during each heartbeat.
The data was then linked with previous medical information fed to the machine learning software. It was then able to learn about human heart’s different attributes, for instance, its shape and structure, eventually predict the risk of failure. The researchers note that their machine learning software is able to predict chances of survival after 1 year with an efficiency of 80%.
Declan O’Regan, who is leading this research, says “this is the first time computers have interpreted heart scans to accurately predict how long patients will live. It could transform the way doctors treat heart patients.”
According to him, a doctor taking the help of new cardiac imaging approaches would have the upper hand while making judgments in comparison to the method currently used in this field.
The involvement of artificial intelligence in medical science is escalating year after year. You might have read another story I covered on Fossbytes, telling IBM Watson’s use for diagnosing a rare leukemia type.
You can check out the research here.
If you have something to add, drop your thoughts in the comments.