Research ArticleCOGNITIVE NEUROSCIENCE

Transcranial brain atlas

See allHide authors and affiliations

Science Advances  05 Sep 2018:
Vol. 4, no. 9, eaar6904
DOI: 10.1126/sciadv.aar6904
  • Fig. 1 Overview of brain atlases and TBAs.

    A brain atlas maps locations in brain space to labels. Because of individual variation, this mapping is probabilistic. A TBA maps locations in scalp space to labels through a probabilistic transcranial mapping from scalp space to brain space.

  • Fig. 2 CPC space.

    (A) CPC space on an example real scalp. (B) BNU projection of CPC space for planar visualization. (C) 10-20 system reference points of 114 participants plotted in CPC space. The tight distributions suggest high consistency between the two systems. (D) Scalp-to-cortex distance (in millimeters), an example data set visualized using CPC space.

  • Fig. 3 Probabilistic transcranial mapping.

    (A) Given one CPC point s = (0.4, 0.6), (B) we identify the corresponding single point on each individual scalp (black dots) using the CPC definition and determine the position of the cortical projection point b (yellow dots) via the balloon inflation model in individual MRI space. We then register these projection points to MNI space. (C) Distribution of the registered projection points is the probabilistic transcranial mapping for s. (D) Variability of the probabilistic transcranial mapping built from our data, quantified by the SD of the MNI coordinates b (in millimeters).

  • Fig. 4 TBA built from the LPBA40.

    Labels are simplified to the major lobes in the upper panels. (A) MPM of TBA114_LPBA_LOBE, showing the probability of the most likely label via BNU projection. (B) MLLM of TBA114_LPBA_LOBE, showing the label with highest likelihood. (C) MPM of TBA114_LPBA. (D) MLLM of TBA114_LPBA. (E) MLLM of TBA114_LPBA rendered using perspective projection on the scalp of a randomly selected participant.

  • Fig. 5 TBA label prediction performance.

    (A) MLLM for TBA92 (the prediction), built from the construction group participants. (B) Label map for an example testing group participant (the ground truth), built from individual sMRI data. (C) Prediction accuracy, as a fraction of testing group participants correctly labeled. (D) Thresholded prediction accuracy map with yellow above 90% and red below 90%.

  • Fig. 6 fNIRS finger-tapping study with TBA-guided probe placement (top) versus 10-20–guided placement (bottom).

    (A) Intended probe locations. Black dots indicate channel locations. Red dots indicate emitters, and blue dots indicate detectors. (B) Consistency of channel locations. Standard deviations across participants of the projected cortical locations for each channel are represented as radii of red discs. (C) Spatial pattern of group-level oxygenated HbO activation during finger-tapping. Color scale depicts t values. (D) ROI coverage. First to seventh columns illustrate the actual ROI coverage of each probe placement. A channel is colored yellow if its projected cortical location was within the precentral gyrus. ROI coverage was summed channel by channel across the subjects, and the group sums are illustrated in the last column. In (B) and (C), the results are represented on the ICBM152_2009b template (41) provided by MRIcroGL (http://www.mccauslandcenter.sc.edu/mricrogl/).

Supplementary Materials

  • Supplementary material for this article is available at http://advances.sciencemag.org/cgi/content/full/4/9/eaar6904/DC1

    Supplementary Results

    Fig. S1. CPC system definition.

    Fig. S2. TBA-based navigation.

    Fig. S3. TBA built from the AAL2 atlas.

    Fig. S4. TBA built from the Talairach Atlas, showing BAs.

    Fig. S5. TBA built from the Craddock400 atlas.

    Fig. S6. Label prediction performance.

    Fig. S7. ROI-wise DICE map.

    Fig. S8. Accuracy maps of cross-racial prediction.

    Table S1. ROI-wise DICE of TBA_LPBA, TBA_AAL, and TBA_BA.

  • Supplementary Materials

    This PDF file includes:

    • Supplementary Results
    • Fig. S1. CPC system definition.
    • Fig. S2. TBA-based navigation.
    • Fig. S3. TBA built from the AAL2 atlas.
    • Fig. S4. TBA built from the Talairach Atlas, showing BAs.
    • Fig. S5. TBA built from the Craddock400 atlas.
    • Fig. S6. Label prediction performance.
    • Fig. S7. ROI-wise DICE map.
    • Fig. S8. Accuracy maps of cross-racial prediction.
    • Table S1. ROI-wise DICE of TBA_LPBA, TBA_AAL, and TBA_BA.

    Download PDF

    Files in this Data Supplement:

Navigate This Article