Magnetic Resonance Imagining (MRI) Magnetic Fields

1. Magnetic Resonance Imagining (MRI) Magnetic Fields • In an externally applied magnetic field, atomic nuclei with an odd number of nucleons (protons + neutrons) precess • = Spin axis wobbles • Rate determined both by properties of the atom • And by the strength of the magnetic field • Hydrogen atoms in a 1.5 Tesla field precess at ~64 MHz • = Radio frequency [rf] range

2. MRI Magnetic Fields Cont’d • After a few seconds in a magnetic field, the spin axes of a small fraction of the relevant nuclei align with each other • = Axes all wobble in same way • Strength of the magnetic field determines how big the fraction is • Spin axes are aligned, but not precessing in phase at this point • Alignment of spin axis of nuclei causes the whole magnetic field generated by the spinning nuclei to precess around the axis of static magnetic field

3. Resonance • An EM pulse with a narrow range of radio frequencies (rf pulse) is applied to the aligned nuclei in the magnetic field • When pulse frequency = precession frequency, nuclei resonate to it • Pushes all of the spinning nuclei into phase with one another • Amplitude of wobble of the whole magnetic field generated by the spinning nuclei increases (= spin axis is pushed farther out) • How far spin axis moves (= flip angle) depends on rf pulse intensity and duration

4. Why is this useful? • Takes characteristic amount of time for nuclei: • To get out of phase (=de-phase) • And to settle back to original amplitude of the wobble in fixed field • Depending on the kinds of molecules the atoms are in • As nuclei settle back into alignment with fixed field, they emit measurable EM energy themselves • Variations in how long it takes the nuclei to de-phase & to settle back to original wobble in fixed field • Can be used to distinguish among different substances

5. How does this make IMAGING possible? • How know where in object EM energy being measured comes from?) • Use gradient magnetic fields (Lauterbur Nobel Prize) • Generate field with gradation in field strength with only a narrow band at 1.5 Tesla • Only hydrogen atoms within that narrow band respond to 64 MHz pulse • So, response must come from atoms within 1.5 T portion of field • Keep moving position of 1.5 T band to localize source of responses to repeated rf pulses (“slices”) • Narrowness of band determines granularity of localization of response

6. Siemens Allegra 3TBiomedical Imaging Center (BIC)

7. Fixed field magnet is always on!

8. Structural MRI • Anatomical scans generally measure hydrogen atoms in water • Since different kinds of tissue have different proportions of water • Typical anatomical scan voxel granularity = 1 x 1 x 7 mm

9. Functional MRI (fMRI) • Substance measured is hemoglobin (iron) in blood • Blood flow increases to active brain regions • Increases more than is usually needed • So ratio of de-oxygenated to oxygenated blood decreases • Oxygenated & de-oxygenated hemoglobin respond differently to magnetic field and rf pulses • Thus, can detect where ratio of oxygenated to de-oxygenated blood changes during/after some event • Takes 6 - 9 seconds for the response to peak • Fastest reliably detectable pre-peak response so far = 2 - 4 sec • Signal strength change very small – generally less than 1% change

12. fMRI Cont’d • Spatial resolution: • Ultimate limit probably spatial specificity of the circulatory system • Worse than structural MRI • Typical functional scan voxels = 3 x 3 x 7 mm • Temporal resolution: • IF blood flow is what’s measured, never going to be faster than seconds • Working on detecting the brief initial decrease in oxygenated blood preceding increased blood flow • Working on imaging other substances

13. Subtractive Logic • Most of the brain is active during most events • Try to isolate regions that are specific to some aspect of the event of interest • So, construct 2 conditions that you believe have just some crucial interesting difference • Treat one as baseline and subtract it from the other, to get rid of all the activity the 2 conditions have in common • And analyze (part of) what’s left

14. Subtractive Logic, Cont’d • Similar logic used in comparing conditions in most other kinds of experiments, too • But there’s been a very unfortunate tendency in the imaging literature so far, • For researchers who don’t have a good understanding of the many ways that different kinds of stimuli and/or tasks and/or situations can differ, • To claim that they’ve located “phonological word processing”, or “irregular morphological inflection processing”, or some such aspect of language processing • When other confounded differences between conditions are equally good candidates (such as plain old difficulty) for explaining the effects