Feasibility of functional MRI on point-of-care MR platforms
Abstract
Magnetic resonance imaging (MRI) has proven to be a clinically valuable tool that can produce anatomical and functional images with improved soft tissue contrast compared to other imaging modalities. There has recently been a surge in low- and mid-field scanners due to hardware developments and innovative acquisition techniques. These compact scanners are accessible, offer reduced siting requirements and can be made operational at a reduced cost.
This thesis aims to implement blood-oxygen-level-dependent (BOLD) resting-state functional MRI (fMRI) at such a mid-field point-of-care scanner. The availability of this technique can be beneficial to get neurological information in cases of traumatic brain injury, stroke, epilepsy, and dementia. This technique was previously not implemented at low- and mid-field since signal-to-noise ratio and the contrast scale with field strength.
Studies were conducted to gauge the performance of an independent component analysis (ICA) based platform (GraphICA) to analyze artificially added noisy resting state functional data previously collected with a 3T scanner. This platform was used in later chapters to preprocess and perform functional connectivity studies with data from a mid-field scanner.
A single echo gradient echo echoplanar imaging (GE-EPI) sequence is typically used for BOLD-based fMRI. Task-based fMRI experiments were performed with this sequence to gauge the feasibility of this technique on a mid-field scanner. Once the feasibility was established, the sequence was further optimized to suit mid-field scanners by considering all the imaging parameters.
Resting-state experiments were conducted with an optimized single echo GE-EPI sequence with reduced dead time on a mid-field scanner. Temporal and image signal-to-noise ratio were calculated for different cortical regions. Along with that, functional connectivity studies and identification of resting-state networks were performed with GraphICA which demonstrated the feasibility of this resting-state fMRI at mid-field. The reliability and repeatability of the identified networks were assessed by comparing the networks identified with 3T data.
Resting-state experiments were conducted with a multi-echo GE-EPI sequence to use the dead time due to long T2* at mid-field effectively. Temporal signal-to-noise was calculated for different cortical regions. Along with that, functional connectivity studies and identification of resting-state networks were performed with GraphICA which demonstrated the feasibility of this resting-state fMRI at mid-field.