Russian scientists have found a new way to use neural networks in medicine. According to researchers from Samara University, applying this technology will increase the MRI scanning accuracy. Results of the study are published in the Computer Optics journal.

According to scientists from Samara University, applying the technology proposed will make it possible to detect interference caused by the patient’s movements right during the MRI scanning session. Now the measurement is complicated by the fact of the patient’s arbitrary movements in the device: they have to be manually found and considered in course of data postprocessing.

Artificial errors, also called MRI artifacts, often make it difficult to decode tomography. They cause the patient’s head to move during the procedure. Interference may cause the scan to be interrupted prematurely due to inaccuracy of the result.

"In the existing technology, motion artifacts are filtered out at the stage of processing MRI experiment data when determining the volume of the brain to the first or average volume of the series using solid state transformation. They include three displacement parameters and three rotation parameters for each amount of the time MRI series”, said Nikita Davydov, Senior Lecturer at the Department of Engineering Cybernetics, an employee of the Artificial Intelligence Institute of Samara University.

According to the expert, work in the area of creating the interference tracking technology during the MRI procedure has been conducted by the University since 2019, and has already been completed and implemented by now. At Samara University, neural networks were trained to detect stepwise artifacts of head movement in fMRI data, and adapted to the set of real data.

“First, the neural network model is trained by using a large amount of synthetic data generated with parameters close to real ones, then by using the small number of sets of real head-movement data corresponding to different people, and after that, the model already works with a small part of the real data corresponding to a specific experiment”, explained Davydov. This approach is called “meta-learning by small amount of data”, and has previously been used in the task of image restoration within the studies conducted by the Artificial Intelligence Institute.

The onward task of the research team is to increase accuracy of the neural network model by creating a procedure for generating synthetic data, closer to real data. The next step for further analysis will be the ability to calculate parameters, such as coordinates, heights, and duration of the anomaly, which will allow filtering out the number of false positives of the classification.

Samara University is a participant in the Russian State University-Support Program “Priority 2030” of the National Project “Science and Universities”.

Source: ria.ru
Image: stablediffusionweb.com