Skip to content

Developer Guide

Dependencies

AnatQC is built upon FreeSurfer, MRIQC, and volumetric navigators software packages. The container is based on Rocky Linux.

Package Version Download
Rocky Linux 8
FreeSurfer 6.0.0
MRIQC 0.15.3
vNav 0.4.0

Pipeline overview

FreeSurfer

FreeSurfer is a full processing stream for MRI imaging data that implements anatomical segmentation, cortical surface reconstruction, registration, and parcellation.

recon-all

AnatQC runs the main FreeSurfer reconstruction pipeline, recon-all, on a single T1w image

Submilimeter resolution

If the input image is submilimeter resolution, the -hires argument will be appended to the recon-all command.

recon-all -sd ${SUBJECTS_DIR} -s ${SUBJECT} -all -i ${NIFTI}

tal_QC_AZS

The tal_QC_AZS command outputs a quality control metric that reflects the accuracy of the Talairach transformation during the -tal-check stage of recon-all

tal_QC_AZS talairach_avi.log

mris_anatomical_stats

The mris_anatomical_stats command outputs metrics such as surface area, gray matter volume, and cortical thickness for a particular hemisphere

mris_anatomical_stats -l lh.cortex.label ${SUBJECT} lh
mris_anatomical_stats -l rh.cortex.label ${SUBJECT} rh

mris_euler_number

The mris_euler_number command produces a quality control metric that reflects the number of surface defects in the unfixed surfaces. A perfectly reconstructed cortical hemisphere surface should be topologically equivalent to a sphere, which should have an expected Euler number of 2. Deviations from this number indicate the presence of topological defects, such as "holes" or self-intersecting "handles"

mris_euler_number lh.orig.nofix
mris_euler_number rh.orig.nofix

wm-anat-snr

The wm-anat-snr command calculates the White Matter anatomical Signal-to-Noise Ratio from a subject's structural image. This metric may be used to identify scans that are affected by excessive head motion, or poor tissue contrast

wm-anat-snr --sd ${SUBJECT} --force

mri_cnr

While wm-anat-snr checks noise level within a single tissue (White Matter), mri_cnr measures Contrast-to-Noise Ratio. This metric reflects how well separated signal intensities are between adjacent tissues. 

mri_cnr ${SUBJECTS_DIR}/${SUBJECT}/surf ${SUBJECTS_DIR/${SUBJECT}/mri/orig.mgz

MRIQC

MRIQC is a well-known tool designed to automate quality control and extraction of image quality metrics from MRI brain images. MRIQC generates an HTML report that contains mosaic views of the brain, segmentaion boundaries, and more.

mriqc

AnatQC runs the main mriqc command at the participant level

mriqc --participant_label ${SUBJECT} --session-id ${SESSION} --run-id ${RUN}
--work-dir ${WORKING_DIR} --verbose-reports --float32 --n_procs 2 --no-sub
${INPUT_DIR} ${OUTPUT_DIR} participant

vNav

Volumetric navigators, or "vNavs", are rapid, low-resolution 3D volume scans slotted into the natural pauses within a longer T1-weighted structural MRI sequence (like MPRAGE). These navigator volumes are used to track and correct for head motion, in real time, without needing additional hardware.

parse_vNav_Motion.py

The parse_vNav_Motion.py command extracts vNav tracking data from the metadata (DICOM headers) of T1- and T2-weighted MRI data, and generates useful head motion metrics and plots. These motion metrics and plots reflect milimeter shifts (RMS and Max) over the timeline of the sequence

parse_vNav_Motion.py --tr ${TR} --rms --max --plot --input-dir ${DICOM_DIR} --output-dir ${OUTPUT_DIR}