FMRI only measures the secondary physiological correlates of neural activity, it is not a direct measure. When nerve cells are active they consume glucose and oxygen. The local response to the glucose/oxygen utilization is an increase in blood flow to regions of increased neural activity. This is this increase in blood flow that is imaged using an MRI scanner.

Nowadays, the relationship between the FMRI signal and the underlying activity is an active area of research. A variety of techniques have been developed to calibrate an individual’s response in order to obtain a quantitative measure of neural activity (i.e., BOLD, ASL, ...)

One of the actual main challenges is to understand of what we are observing, and seeing, and measuring, and wondering about. I think that challenges related to signal quality, scanner improvements and all of that should remain as secondary issues until the community explains how fMRI works.

I dont understand your last question about neuroscience and machine learning... is it related to MRI?

Regards.

Dear Martín Martínez Villar

Firstly I want to know how we use both MRI and fMRI together to represent the brain activity.

Machine learning give solution about how to learn the knowledge from any data by machine, in somehow some machine learning algorithms emulate the brain activity in observation, vision, recognition. I think there is a reciprocal relationship between Machine learning and neuroscience. In other word there is a lot of data that produced by neuroscience about brain, and have big challenges in processing, or in representing the hidden knowledge about brain activity.

What the other kinds of data captured from brain?

Is the fMRI related to time? "Time series data"

Cheers

Hi Osama,

MRI employes sequences programmed on the scanner to acquire the images (in BOLD-fMRI it is normally the T2*). It represents a time series with some particularities.

I think you refer to this kind of issues:

Machine Learning for Clinical Diagnosis from Functional Magnetic Resonance

Imaging (http://citeseerx.ist.psu.edu/viewdoc/download?doi=10.1.1.150.8160&rep=rep1&type=pdf)

Machine Learning with Brain Graphs (http://miplab.epfl.ch/pub/richiardi1301.pdf)

Brain decoding based on functional magnetic resonance imaging using machine learning: a comparison study (http://espace.library.uq.edu.au/view/UQ:312376)

I think that the first document will be useful for your purpose, hope it helps you.

Best,

Yes, fMRI analyse time series data. The conventional approach is to apply general linear modelling to all voxel time series and check which voxel(s) have a significant correlation with the contrast-of-interest. This approach is univariate in nature, i.e. the analysis is performed on voxel-by-voxel basis.

However, there is a new trend to consider signals from multiple voxels as a pattern and feed them into machine learning algorithms. Using machine learning algorithms on fMRI BOLD or EEG/EMG signals is a very new and active field in the imaging community. People had successfully made classification for clinical diagnosis, predict active brain states using activation paradigm (brain decoding), and even develop brain-machine interface with real-time fMRI.

Just a few seminal papers you probably would interested in:

http://www.sciencemag.org/content/293/5539/2425

http://www.nature.com/neuro/journal/v8/n5/full/nn1444.html

Soon, C.S., Brass, M., Heinze, H.J.& Haynes, J.D. (2008). Unconscious determinants of free decisions in the human brain. Nature Neuroscience 11, 543-5.

Reconstructing Visual Experiences From Brain Activity Evoked by Natural Movies

Shinji Nishimoto, An T. Vu, Thomas Naselaris, Yuval Benjamini, Bin Yu & Jack L. Gallant (Current Biology 2011)

Good day ,

As Far as I read ,fmRI provides good spatial resolution of brain activity but have a poor temporal resolution ( seconds).So It will not be suited for real - time application.

For Use in Brain Computer Interface,The Area In Which I am working on Now ,

This might not be of great use for daily usage due to expense of the machine itself not affordable for end users and due to its lack of portability /mobility.

But as far spatial resolution of brain activity is concerned ,it very nice for offline processing of brain images.

I read somewhere that real time fmRI has been developped (not very sure).

may be you can check as well the advatanges and disadvantage of hemodynamic response (fmRI,..) vs electric response (EEG,LFPs,Mutilple Units,...)

But I believe the challenges may also relate to the direction and application you are looking at (real time applications,....).

I hope this helps.

Thanks.

I believe Martín Martínez Villar mentioned a very important information regarding hemodynamic response (fmRI) which I read about concerning how they correlate with brain activity.

May be you can do extra reading on it as well.

Thanks