1. Neuroscience
Download icon

Multi-contrast anatomical subcortical structures parcellation

  1. Pierre-Louis Bazin  Is a corresponding author
  2. Anneke Alkemade
  3. Martijn J Mulder
  4. Amanda G Henry
  5. Birte U Forstmann
  1. Integrative Model-based Cognitive Neuroscience research unit, University of Amsterdam, Netherlands
  2. Max-Planck Institute for Human Cognitive and Brain Sciences, Germany
  3. Psychology Department, Utrecht University, Netherlands
  4. Faculty of Archaeology, Leiden University, Netherlands
Research Article
Cite this article as: eLife 2020;9:e59430 doi: 10.7554/eLife.59430
11 figures, 4 tables and 1 additional file

Figures

The 17 subcortical structures currently included in the parcellation algorithm in axial (A), sagittal (B), and coronal (C) views.
The MASSP parcellation pipeline.

Atlas priors for interfaces between structures are combined to the MRI data, regularized via probability diffusion and topology correction, and the final structure posteriors are jointly estimated by region growing.

Leave-one-out validation of the structures parcellated by MASSP, compared to the human rater with most neuroanatomical expertise.

Scores for the left and right side are computed separately and then combined into box-and-whisker plots.

Inter-rater variability for the human expert raters.

Scores for the left and right side are computed separately and then combined into box-and-whisker plots.

Parcellation with Freesurfer (top, on T1w image) and MASSP (bottom, on T2w image) on Human Connectome Project data.

MASSP priors were not derived from the contrasts, but transferred via a spatial mapping of the quantitative MRI intensities from AHEAD subjects.

MASSP parcellation scores as a function of increasing number of subjects included in the atlas.

Scores for the left and right side are computed separately and then combined into box-and-whisker plots.

MASSP parcellation scores over the lifespan.

Each matrix show the average Dice overlap (top), dilated Dice overlap (middle), and average surface distance (bottom) for using one age group as prior (’train’) to parcellate another age group (’test’).

Regression of volume (log scale), structure thickness, R1, R2*, and QSM MRI parameters estimated using manual delineations versus MASSP automated parcellations.

Circles show individual data points, linear regression is indicated by a straight line, and 95% confidence interval is given as the shaded area. Pearson correlation coefficients are indicated when significant (p-value<0.01).

MP2RAGEME maps and delineations: quantitative R1 (left), quantitative R2* (middle), QSM (right).

Manual delineations for the 17 structures of interest are overlaid on all images.

Anatomical interface (A) and skeleton (B) priors derived from the 10 manually delineated subjects.
Successive parcellation results: (A) voxel-wise posteriors and parcellation, (B) diffused posteriors and parcellation, (C) topology-corrected posteriors and final region-growing parcellation.

Tables

Table 1
Mean overlap and distance measures for the leave-one-out validation.
StrSTNSNRNGPiGPeThaLV3V4VAmgicVTAfxPAGPPNCl
Dice overlap
MASSP0.8930.6480.8050.8700.7020.8000.8670.8490.7410.8690.7230.7450.5700.5270.6410.4960.485
Manual0.8970.8000.8410.8750.7620.8130.8770.9070.7970.8820.8660.7320.5740.8230.7910.6650.727
Ratio0.9950.8110.9570.9960.9250.9870.9890.9360.9360.9880.8361.0200.9940.6410.8140.7540.664
Dilated overlap
MASSP0.9820.9190.9770.9910.9090.9560.9700.9290.8900.9510.8910.9150.8630.7560.8970.7950.789
Manual0.9870.9880.9850.9950.9530.9720.9700.9670.9440.9610.9780.9240.8180.9570.9600.9100.914
Ratio0.9950.9300.9920.9950.9550.9841.0000.9610.9460.9910.9110.9921.0590.7900.9350.8790.863
Average surface distance
MASSP0.7500.9110.6760.4911.1400.8631.0582.6900.9940.8171.4761.2751.0742.9500.9551.4841.685
Manual0.7230.5080.5710.4820.9020.8040.9710.6150.6370.6710.7791.0451.2040.7030.5550.8010.670
Ratio0.9710.5900.8520.9960.8610.9430.9160.2770.6621.0200.5530.8341.1610.2870.6100.6190.465
Table 2
Comparison with multi-atlas registration, Freesurfer, and FSL FIRST.
StrSTNSNRNGPiGPeThaLV3V4VAmgIcVTAFxPAGPPNCl
Dice overlap
MASSP0.8930.6480.8050.8700.7020.8000.8670.8490.7410.8690.7230.7450.5700.5270.6410.4960.485
Multi-atlas0.8550.6620.7600.8200.7420.7960.8590.7340.6600.6910.7610.7180.6260.4780.6740.5390.398
Freesurfer0.8760.7780.8380.8580.4300.7690.692
FSL FIRST0.8750.8130.8390.653
Dilated overlap
MASSP0.9820.9190.9770.9910.9090.9560.9700.9290.8900.9510.8910.9150.8630.7560.8970.7950.789
Multi-atlas0.9760.9380.9680.9890.9470.9680.9700.9200.9200.9080.9210.9390.9240.7980.9430.8710.811
Freesurfer0.9750.9220.9460.9740.5620.9110.857
FSL FIRST0.9760.9460.9500.843
Average surface distance
MASSP0.7500.9110.6760.4911.1400.8631.0582.6900.9940.8171.4761.2751.0742.9500.9551.4841.685
Multi-atlas0.9610.8910.8580.6750.9920.8821.0831.4170.9321.2491.3591.1290.8131.3620.7941.0551.273
Freesurfer0.7701.2111.4050.6854.0711.3611.749
FSL FIRST0.8671.1431.6751.746
Volume bias
MASSP0.0410.017-0.0380.0070.0660.0890.0400.04700.1210.0470.0780.1830.032-0.0160.0260.0090.023
Multi-atlas0.020-0.087-0.0090.0310.0090.0140.020-0.003-0.007-0.092-0.0380.055-0.067-0.264-0.090-0.269-0.376
Freesurfer0.0170.0870.1630.122-0.5510.3510.468
FSL FIRST-0.100-0.0210.165-0.249
Table 3
Test-retest comparison with Freesurfer on Human Connectome Project data.
StrSTNSNRNGPiGPeThaLV3V4VAmgicVTAfxPAGPPNCl
Dice overlap
MASSP test-retest0.9140.7010.8180.8290.7910.8590.9280.8810.8370.8700.8660.8600.7380.7740.7140.7130.785
Freesurfer test-retest0.8980.7700.9190.8940.8420.8490.852
MASSP – Freesurfer0.8760.7780.8380.8580.4300.7690.692
Dilated overlap
MASSP test-retest0.9870.9390.9770.9780.9630.9770.9900.9800.9790.9860.9810.9730.9690.9610.9650.9720.966
Freesurfer test-retest0.9860.9260.9860.9890.9720.9750.978
MASSP – Freesurfer0.9540.7880.9190.9340.4350.9010.866
Average surface distance
MASSP test-retest0.5130.5280.4670.4610.5320.5080.4880.5090.3910.4190.5330.5360.4310.4640.4280.4020.433
Freesurfer test-retest0.8760.7780.8380.8580.4300.7690.692
MASSP – Freesurfer0.9761.6731.6051.9465.6991.4281.478
Table 4
Mean volume and quantitative MRI values for each age group.
Age
StrSTNSNRNGPiGPeThaLV3V4VAmgicVTAfxPAGPPNCl
Volume (mm3)
18-401065611856625356713667112752418951391131543352541632250193843
41-601057212458325658614037492885020241408136344952641808255195830
61-801073413058426058613977463914220231407132144072721910259192829
Thickness (mm)
18-405.941.892.554.643.093.568.314.272.774.034.814.061.692.061.781.921.79
41-605.471.862.664.582.963.418.285.082.953.894.854.191.761.961.841.861.80
61-805.221.832.604.112.923.228.284.903.184.064.734.191.801.971.901.951.82
qR1 (Hz)
18-400.6470.9490.8570.9280.8680.8500.7610.3320.3460.2740.5460.9060.8190.7140.6540.7790.650
41-600.6620.9680.8930.9390.8790.8560.7580.2780.3150.2690.5590.9040.8330.6710.6530.7710.664
61-800.6480.9520.8820.9030.8600.8300.7430.2730.3000.2700.5520.8840.8140.6380.6470.7640.669
qR2* (Hz)
18-4043.867.167.863.275.279.638.114.718.99.025.536.839.237.425.932.732.6
41-6050.474.174.177.180.287.940.38.412.411.728.138.742.837.428.033.436.9
61-8051.877.072.573.877.887.040.18.510.212.030.139.652.635.728.434.235.4
QSM (ppm)
18-400.03290.06090.07380.07170.10150.11500.01380.01300.01000.02790.0036−0.02340.02410.00790.01190.0135−0.0122
41-600.04000.06470.07130.08290.09840.12410.01340.01150.00250.02340.0085−0.02260.02010.00790.00890.0099−0.0110
61-800.04110.07050.06100.07380.09250.12490.00640.0089−0.00340.02360.0061−0.02430.01770.01000.00390.0096−0.0091

Additional files

Download links

A two-part list of links to download the article, or parts of the article, in various formats.

Downloads (link to download the article as PDF)

Download citations (links to download the citations from this article in formats compatible with various reference manager tools)

Open citations (links to open the citations from this article in various online reference manager services)