Overcoming barriers for using rs-fMRI for diagnosis of brain disorders
The capability of rs-fMRI to measure blood oxygen level dependent (BOLD) changes helped rs-fMRI adoption in clinical research. rs-fMRI is useful in understanding neural processing systems at a macro level and while its use in neuroimaging clinical studies is quite common, rs-fMRIs are yet to see widespread adoption in clinical medicine.
O’Connor and Zeffiro’s (2019) study of 71 neuroradiology professionals discovered the reasons behind this trend.Their study showed that fMRI noise, lack of robust single-subject analysis, unreliable mapping of functional connectivity and dearth of training opportunities for neuroradiologists are barriers to the adoption of rs-fMRI in clinical settings.
Group vs Individual scans
In clinical research, the scans of multiple healthy individuals and patients are averaged and then compared. In clinical medicine, it is difficult to obtain reliable data for a single individual due to lack of a comparison group.There is a need for better methodology to derive rs-fMRI maps at the individual level. With advancements in technology, such as higher powered tesla machines, quicker image acquisition and multiple channel coils this is now possible. BrainSightAI patented data-processing pipeline has achieved sensitivity improvements that can help in individual level analysis
Collecting data vs processing data
Neurologists say that it is easy to collect data through rs-fMRI but processing it proves to be cumbersome, making it unsuitable for busy medical clinicians. Voxelbox, BrainSightAI’s proprietary artificial intelligence and machine learning software reduces processing time to under 24 hours making it convenient for neurologists to use rs-fMRI in their clinical investigation.
Another common problem is physiological noise including head movement, heart pulsation and respiration. These physiological aspects can be detrimental to detecting regional neural modulations. To overcome this challenge, clinicians use a movie loop to repeat scans when they detect excessive motion. Thus, motion correction with precise body alignment is necessary and part of the pre-processing aspect of fMRIs. Similarly, cardiac pulsations and movement due to respiration can be corrected using band-pass filtering as it removes frequencies beyond 0.01 Hz - 0.2 Hz. BrainSightAI's VoxelBox technology processes all these different types of noises.
Currently there is no standard protocol when it comes to rs-fMRI acquisition and analysis. This is the major reason rs-fMRIs have been limited to clinical research as a long research pipeline is needed to develop a pipeline for a particular challenge. 7 years of thorough research by Dr. Rimjhim Agarwal (now co-founder and CTO BrainSightAI) led to the development of standardised and automated solutions making it applicable in medical settings.
At Voxelbox, clinical researchers have access to a full-stack team of neuroscientists, artificial intelligence experts, neurologists and grant writing professionals to increase the pace of research advancement. Voxelbox’s commitment to research ensures that the research to application turnaround is quick.
Voxelbox aims to tap the high reproducibility of resting state networks within-subjects to provide a biological basis of brain disorders and aid clinical management of disease progression in patients.
O'Connor, E. E., & Zeffiro, T. A. (2019). Why is clinical fMRI in a resting state?. Frontiers in neurology, 10, 420