Basic concepts:

• sulci
• gyri
• "landmarks"
• lobes (parietal, temporal, occipital, frontal)
• volumes, slices
• surfaces
• axial (transverse), sagittal, coronal, anterior-posterior (rostral-caudal), inferior-superior (ventral-dorsal), left-right, lateral-medial
• T1 volumes are most common to visualize. FreeSurfer can reconstruct cortical surfaces from a single T1 volume.
• ribbon
• FreeSurfer surfaces (native, pial, white, inflated, sphere, sphere.reg)
• atlas - e.g. Glasser (HCP MMP1.0), Wang (Kastner) 2015
• thickness, curvature, sulc
• flatmap - picture of entire cortical surface where its 3D structure is gone (it's been completely flattened)
• MNI - "volume-based" anatomical template
• accuracy and precision is limited, so be careful
• there are different MNI templates vs. the MNI space
• fsaverage - surface-based "curvature-based" anatomical template
• accuracy should be better than volume-based, but subjects aren't fully completed "aligned"
• 163842 vertices per hemisphere in the "canonical" fsaverage surfaces
• Talairach is another space, but it's old and obsolete
• tricky and complicated issues:
• anatomy vs. functionally-defined regions
• variability in conventions in how you define and label an area
• real inter-individual variability
• Spatial scales - 1mm, 2mm, 5mm, 10mm - how much you care depends on the scale you care about. Also, the resolution of your fMRI (e.g. 2-3mm isotropic acquisition) measurements will set the stage for how much you care.
• ROIs - they are hard to do well. partly because there is real variability. arguably the most accurate approach is define ROIs precisely in individual subjects
• 'Functionally-defined' regions to some degree follow the anatomy (sulci/gyri), but not completely well.
• Functional hyperalignment - are you willing to spatially "scramble" your voxels?

Checklist for brain maps:

• indicate sulci/gyri landmarks?
• show the colormap and the color range
• indicate LH/RH
• provide an alternative view (inflated instead of just flat?)
• show the underlying cortical curvature
• provide outlines indicating locations of specific ROIs?
• indicate approximate directions (e.g. dorsal, ventral, etc.)

Figures and other resources:

•  fmribasics  (see materials/20151201 neurostatistics fMRI Part 1 (Main).pdf)
•  one-offs/neuroanatomy/ 
•  nsddata/inspections/ 

Literature:

•  Where Is Human V4? Predicting the Location of hV4 and VO1 from Cortical Folding 
•  Mapping hV4 and ventral occipital cortex: The venous eclipse 
•  Not one extrastriate body area: Using anatomical landmarks, hMT+, and visual field maps to parcellate limb-selective activations in human lateral occipitotemporal cortex 
•  The mid-fusiform sulcus: A landmark identifying both cytoarchitectonic and functional divisions of human ventral temporal cortex 
•  Probabilistic Maps of Visual Topography in Human Cortex 
•  Visual Field Maps in Human Cortex 
•  Temporal Processing Capacity in High-Level Visual Cortex Is Domain Specific 
•  101 labeled brain images and a consistent human cortical labeling protocol 
•  Identification of the ventral occipital visual field maps in the human brain 
•  Overlooked tertiary sulci serve as a meso-scale link between microstructural and functional properties of human lateral prefrontal cortex 

Notes from Jon Winawer: