fMRI_Intro

HOW FMRI WORKS IN ONE SLIDE!

Blood Oxygen Level Dependent (BOLD) Contrast

• Oxyhemoglobin: diamagnetic; Deoxyhemoglobin: paramagnetic

• Paramagnetic  Susceptibility  T2* signal loss (e.g. DSC perfusion)

• Brain activity  demand for O2  regional CBF 

(“Neurovascular coupling”  “Hemodynamic response”)

• CBF  Oxy:Deoxy Hgb (O2 supplied in excess)  T2* signal 

• Signal change small (~3%) – not visible – need stats to ID active regions

• Acquire T2* image time series while activity repeatedly cycles on/off

• Examine T2* signal-vs-time voxel by voxel – look for expected cycling

• Cycling present or not? Question of statistical likelihood (with a p-value!)

• Generate topographical map of likelihood and apply threshold (e.g. p<0.05)

• Voxels exceeding threshold fill in with color, make the rest invisible

• Overlay onto anatomy

QA in fMRI

• Things you cannot control (but need to check):

• Did the patient attend to the task and perform it adequately?

• Was head motion excessive or highly synchronized with the task? (False +)

• Is neurovascular coupling intact? (no BOLD effect without it! False –)

• Are there susceptibility artifacts in the region of interest? (False –)

• Things you do control:

• Is the statistical threshold set too high or too low? (More art than science)

• Is the anatomical underlay appropriate for the lesion? (T1, T2, T2-FLAIR)

• Is the anatomical underlay windowed and leveled appropriately?

• Is the DTI opacity optimized for both lesion and tract visibility?

• All 3 planes sent to PACS?