Neural Network Model for Estimation of the Induced Electric Field During Transcranial Magnetic Stimulation

• Published in: IEEE transactions on magnetics Vol. 58; no. 2; pp. 1 - 5
• Main Authors: Afuwape, Oluwaponmile F., Olafasakin, Olumide O., Jiles, David C.
• Format: Journal Article
• Language: English
• Published: New York IEEE 01.02.2022; The Institute of Electrical and Electronics Engineers, Inc. (IEEE)
• Subjects: Artificial neural networks; Brain; Brain modeling; Coils; Computational modeling; Deep convolutional neural network (deep CNN); Electric fields; Electromagnetic induction; Estimation; Finite element method; induced electric field (<italic xmlns:ali="http://www.niso.org/schemas/ali/1.0/" xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance">E -field); Magnetic heads; Magnetic induction; Magnetic resonance imaging; Magnetism; Mathematical analysis; MRI scans; Neural networks; Neurons; Scalp; Solid modeling; Three dimensional models; Transcranial magnetic stimulation; transcranial magnetic stimulation (TMS)
• ISSN: 0018-9464; 1941-0069
• DOI: 10.1109/TMAG.2021.3086761

Transcranial magnetic stimulation (TMS) is a noninvasive neuromodulation technique with applications in brain mapping and effective neuronal connectivity. In its repetitive mode, it is used for the treatment of neurological and psychiatric disorders. It functions on the principle of electromagnetic induction, where generated magnetic fields (<inline-formula> <tex-math notation="LaTeX">H </tex-math></inline-formula>-field) induce electric field (<inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-field) that stimulates the brain's neurons. With TMS studies, accurate estimation of the induced <inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-field is usually necessary. However, this requires a lot of processes, including the 3-D head model generation from magnetic resonance imaging (MRI) scans using the SimNIBS software and finite element analysis to calculate the induced <inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-field. These processes are time-consuming and computationally expensive. In addition, with each head model's uniqueness, outcomes cannot be generalized across a particular population as the intensity of <inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-field is subject-specific. In this research, the authors develop deep convolutional neural network (deep CNN) models to determine the intensity of the induced <inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-field directly from the patient's MRI scan and across different coil types. We trained CNN models from anatomically realistic head models and across 16 coil types to predict the induced <inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-field in the brain and scalp (<inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-Max brain and scalp), and the volume of stimulation of the brain and scalp (<inline-formula> <tex-math notation="LaTeX">V </tex-math></inline-formula>-half brain and scalp) from T1-weighted MRI scans. Using a deep CNN model, the processing time for estimating the induced <inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-field is significantly reduced, which is helpful both to clinicians and researchers as the need to create subject-specific anatomical head structures is eliminated. Also, there will be no need for additional stimulation sessions with the different coil types for TMS patients as the deep CNN model can predict the outcome from each coil type. The other advantages of the deep CNN model are that the <inline-formula> <tex-math notation="LaTeX">E </tex-math></inline-formula>-field from the different coil types can be compared simultaneously.