Manual MRI parcellation of the frontal lobe

Abstract

The ability to examine associations between neuropsychiatric conditions and functionally relevant frontal lobe sub-regions is a fundamental goal in neuropsychiatry, but methods for identifying frontal sub-regions in MR (magnetic resonance) images are not well established. Prior published techniques have principally defined gyral regions that do not necessarily correspond to known functional divisions. We present a method in which sulcal-gyral landmarks are used to manually delimit functionally relevant regions within the frontal lobe: primary motor cortex, anterior cingulate, deep white matter, premotor cortex regions (supplementary motor complex (SMC), frontal eye field and lateral premotor cortex) and prefrontal cortex (PFC) regions (medial PFC, dorsolateral PFC (DLPFC), inferior PFC, lateral orbitofrontal cortex (OFC) and medial OFC). Feasibility was tested by applying the protocol to brain MR data from 15 boys with attention-deficit/hyperactivity disorder (ADHD) and 15 typically developing controls, 8-12 years old. Intra- and inter-rater intraclass correlation coefficients were calculated using parcellation volumes from a subset of that group. Inter-rater results for the 22 hemisphere specific sub-regions ranged from 0.724 to 0.997, with all but seven values above 0.9. Boys with ADHD showed significantly smaller left hemisphere SMC and DLPFC volumes after normalization for total cerebral volume. These findings support the method as a reliable and valid technique for parcellating the frontal lobe into functionally relevant sub-regions.

Figure 1

Parcellated left frontal lobe of an 11.5 year-old boy on a) lateral, b) medial, c) dorsal and d) ventral surfaces. The deep white matter region is not visible on the surface of the brain; a sample DWM can be seen in Figure 2.

Figure 2

A sample frontal lobe parcellation in a series of axial cross-sections. Images from left to right move from more inferior to more superior positions in the brain.

Figure 3

Plots of volume after normalization for TCV, grouped by diagnosis for left DLPFC, left SMC WM and left DLPFC WM. Horizontal bars indicate the mean and standard error for each group. Findings showed significantly reduced volume in ADHD in all three regions, with the effect of outliers mitigated by the use of non-parametric analyses.