Thesis defense Rasim Boyacioglu (Donders Series 158)

8 May 2014

Promotor: Prof.dr. D. Norris, copromotor: dr. M. Barth

On the application of ultra-fast fMRI and high resolution multiband fMRI at high static field strengths

In his thesis, Rasim Boyacioglu has investigated different ways of acceleration of functional MRI (fMRI) data acquisition. fMRI looks at signal changes in the brain triggered by an experimental condition. If each 3D image of the brain (one acquired in ~2-3 s) is thought as a snapshot, faster acquisition would correspond to shooting a video with a higher frame rate. This essentially results in more observation points of the brain function, thus better statistics. Also, communication between different brain regions happens in the millisecond range, so one needs to get faster to understand the connectivity and causality (which region is driving which one) of the brain. One major drawback of fMRI is the delay between the neural firing (onset of the activation) and the observed peak in the fMRI signal. The delay varies between people and brain regions. Due to this fact skeptics of fMRI would oppose using fMRI for investigation of any temporal phenomena. However, with the increasing acquisition speed, we would have a better chance of depicting the nature of this delay and then incorporating this information to our models. With this motivation, in the first part of his PhD, Rasim made improvements on an ultra-fast fMRI technique, called Generalized Inverse Imaging (GIN), which allowed scanning the brain with 50 ms resolution at 3 Tesla. This technique is an extreme case of parallel imaging (PI). As the name suggests in PI the brain is scanned by many coil channels simultaneously. Then the rich and different spatial information provided by many coil channels is exploited by leaving out some of the data collection in a controlled fashion. The missing information is recovered through different sensitivities of the coil channels. In the second part of his PhD he was involved in studies related to faster fMRI data acquisition with multiband (MB) imaging at higher magnetic fields (7 Tesla). Instead of skipping some of the data acquisition steps like the previous PI techniques, MB imaging collects data from multiple slices of the brain simultaneously and then separates them. However, in this case the limitation is the increased power deposition as multiple slices have to be excited simultaneously. One possible solution is to design special pulses whose power do not increase with the number of slices such as the PINS pulses developed at the Donders Institute. Rasim and his colleagues has tested the performance of two MB pulse sequences with PINS pulses and compared their performances with a color-word Stroop task and a resting state fMRI study.