Preprocessing

Data preprocessing has a significant impact on the quality of the reconstructed volume in slice-to-volume reconstruciton. The NeSVoR toolkit provides the following preprocessing steps:

Preprocessing in NeSVoR toolkit

The preprocessing can be performed by using a standalone command (i.e., segment-stack, correct-bias-field, and assess). They can also be integrated into the reconstruction pipeline (i.e., reconstruct and svr) by setting the corresponding flags.

Brain masking

We integrate a deep learning based fetal brain segmentation model (MONAIfbs) into our pipeline to extract the fetal brain ROI from each input image. The segment-stack command generates brain mask for each input stack as follows.

nesvor segment-stack \
    --input-stacks stack-1.nii.gz ... stack-N.nii.gz \
    --output-stack-masks mask-1.nii.gz ... mask-N.nii.gz \

You may also perform brain segmentation in the reconstruct command by setting --segmentation.

Bias field correction

We also provide a wrapper of the N4 algorithm in SimpleITK for bias field correction. The correct-bias-field command correct the bias field in each input stack and output the corrected stacks.

nesvor correct-bias-field \
    --input-stacks stack-1.nii.gz ... stack-N.nii.gz \
    --stack-masks mask-1.nii.gz ... mask-N.nii.gz \
    --output-corrected-stacks corrected-stack-1.nii.gz ... corrected-stack-N.nii.gz

You may perform bias field correction in the reconstruct command by setting --bias-field-correction.

Stack quality assessment

The assess command evalutes the image quality / motion of input stacks. This information can be used to find a template stack with the best quality or filter out low-quality data. An example is as follows.

nesvor assess \
    --input-stacks stack-1.nii.gz ... stack-N.nii.gz \
    --stack-masks mask-1.nii.gz ... mask-N.nii.gz \
    --metric <metric> \
    --output-json result.json

The provided metrics are listed here.