Deep learning for emulating bio-electromagnetic interactions in MRI
Magnetic resonance imaging (MRI) equipment requires properly designed coils to obtain high-resolution images, which should also obey safety regulations to avoid thermal burn injuries of the human body during MRI scanning. However, generated heat and induced fields in MRI vary from patient to patient...
Saved in:
Main Author: | |
---|---|
Other Authors: | |
Format: | Final Year Project |
Language: | English |
Published: |
Nanyang Technological University
2021
|
Subjects: | |
Online Access: | https://hdl.handle.net/10356/149839 |
Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
Institution: | Nanyang Technological University |
Language: | English |
id |
sg-ntu-dr.10356-149839 |
---|---|
record_format |
dspace |
spelling |
sg-ntu-dr.10356-1498392023-07-07T17:49:50Z Deep learning for emulating bio-electromagnetic interactions in MRI Lim, Shannon Shi Mei Abdulkadir C. Yucel School of Electrical and Electronic Engineering acyucel@ntu.edu.sg Engineering::Electrical and electronic engineering::Antennas, wave guides, microwaves, radar, radio Magnetic resonance imaging (MRI) equipment requires properly designed coils to obtain high-resolution images, which should also obey safety regulations to avoid thermal burn injuries of the human body during MRI scanning. However, generated heat and induced fields in MRI vary from patient to patient due to the difference in tissue conductivities. Therefore, to ensure designed coils are safe for everybody, this project aims to quantify the uncertainty from human head tissue conductivities during MRI. During uncertainty quantification (UQ), emulation of MRI is indispensable and will be often required enormous times. As emulators for bio-electromagnetic interactions in MRI, two deep neural networks are implemented in this final year project achieving mean absolute percentage error (MAPE) of 4.46% and 2.90%, respectively. Compared with one currently available MRI simulator, MARIE, the time consumed for one simulation reduces from 10 minutes to less than 1 second, which will significantly improve the efficiency of UQ. DNNs developed guarantee quicker analysis of electric fields based on tissue conductivities while including uncertainties in EM analysis allowing designed coils to be safe for everybody. Bachelor of Engineering (Electrical and Electronic Engineering) 2021-06-09T02:30:10Z 2021-06-09T02:30:10Z 2021 Final Year Project (FYP) Lim, S. S. M. (2021). Deep learning for emulating bio-electromagnetic interactions in MRI. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/149839 https://hdl.handle.net/10356/149839 en A3001-201 application/pdf Nanyang Technological University |
institution |
Nanyang Technological University |
building |
NTU Library |
continent |
Asia |
country |
Singapore Singapore |
content_provider |
NTU Library |
collection |
DR-NTU |
language |
English |
topic |
Engineering::Electrical and electronic engineering::Antennas, wave guides, microwaves, radar, radio |
spellingShingle |
Engineering::Electrical and electronic engineering::Antennas, wave guides, microwaves, radar, radio Lim, Shannon Shi Mei Deep learning for emulating bio-electromagnetic interactions in MRI |
description |
Magnetic resonance imaging (MRI) equipment requires properly designed coils to obtain high-resolution images, which should also obey safety regulations to avoid thermal burn injuries of the human body during MRI scanning. However, generated heat and induced fields in MRI vary from patient to patient due to the difference in tissue conductivities. Therefore, to ensure designed coils are safe for everybody, this project aims to quantify the uncertainty from human head tissue conductivities during MRI. During uncertainty quantification (UQ), emulation of MRI is indispensable and will be often required enormous times. As emulators for bio-electromagnetic interactions in MRI, two deep neural networks are implemented in this final year project achieving mean absolute percentage error (MAPE) of 4.46% and 2.90%, respectively. Compared with one currently available MRI simulator, MARIE, the time consumed for one simulation reduces from 10 minutes to less than 1 second, which will significantly improve the efficiency of UQ. DNNs developed guarantee quicker analysis of electric fields based on tissue conductivities while including uncertainties in EM analysis allowing designed coils to be safe for everybody. |
author2 |
Abdulkadir C. Yucel |
author_facet |
Abdulkadir C. Yucel Lim, Shannon Shi Mei |
format |
Final Year Project |
author |
Lim, Shannon Shi Mei |
author_sort |
Lim, Shannon Shi Mei |
title |
Deep learning for emulating bio-electromagnetic interactions in MRI |
title_short |
Deep learning for emulating bio-electromagnetic interactions in MRI |
title_full |
Deep learning for emulating bio-electromagnetic interactions in MRI |
title_fullStr |
Deep learning for emulating bio-electromagnetic interactions in MRI |
title_full_unstemmed |
Deep learning for emulating bio-electromagnetic interactions in MRI |
title_sort |
deep learning for emulating bio-electromagnetic interactions in mri |
publisher |
Nanyang Technological University |
publishDate |
2021 |
url |
https://hdl.handle.net/10356/149839 |
_version_ |
1772827861918416896 |