Evaluating the performance of 3-tissue constrained spherical deconvolution pipelines for within-tumor tractography

Abstract

The use of diffusion MRI (dMRI) for assisting in the planning of neurosurgery has become increasingly common practice, allowing to non-invasively map white matter pathways via tractography techniques. Limitations of earlier pipelines based on the diffusion tensor imaging (DTI) model have since been revealed and improvements were made possible by constrained spherical deconvolution (CSD) pipelines. CSD allows to resolve a full white matter (WM) fiber orientation distribution (FOD), which can describe so-called "crossing fibers": complex local geometries of WM tracts, which DTI fails to model. This was found to have a profound impact on tractography results, with substantial implications for presurgical decision making and planning. More recently, CSD itself has been extended to allow for modeling of other tissue compartments in addition to the WM FOD, typically resulting in a 3-tissue CSD model. It seems likely this may improve the capability to resolve WM FODs in the presence of infiltrating tumor tissue. In this work, we evaluated the performance of 3-tissue CSD pipelines, with a focus on within-tumor tractography. We found that a technique named single-shell 3-tissue CSD (SS3T-CSD) successfully allowed tractography within infiltrating gliomas, without increasing existing single-shell dMRI acquisition requirements.1 The term "tissue" is used in an abstract sense in context of these multi-tissue CSD techniques: it refers in general to extra signal compartments in the model, even e.g. (cerebrospinal) fluid.

Test-retest reliability and long-term stability of 3-tissue constrained spherical deconvolution methods for analyzing diffusion MRI data

“…Finally, we upsampled the preprocessed data to 1.3⨉1.3⨉1.3mm 3 isotropic voxels (Greenspan, 2008;Kuklisova-Murgasova et al, 2012;Bastiani et al, 2019). These preprocessing steps are largely similar to those used in other recently published works (Bastiani et al, 2019;Pietsch et al, 2019;Mito et al, 2019;Aerts et al, 2019). Brain masks were obtained for all subjects by performing a recursive application of the Brain Extraction Tool (Avants et al, 2014).…”

“…Even for WM tractography in cases of infiltration by pathological tissues, the 3-tissue CSD approach can provide direct benefits in terms of recovering healthy WM structures, e.g. in infiltrating tumors (Aerts et al 2019).…”

Modeling brain dynamics after tumor resection using The Virtual Brain

“…Based on the resulting white matter fiber orientation distributions (FODs), probabilistic streamlines tractography was performed, using an FOD amplitude threshold of 0.07 (Tournier, Calamante, & Connelly, 2010). Although white matter FODs could be estimated using SS3T-CSD within regions infiltrated by a tumor (Aerts et al, 2019), resulting white matter FOD amplitudes were substantially smaller in tumor regions compared to the rest of the brain. While this likely reflects the smaller portion of space taken up by axons (due to infiltrating tumor tissue) and/or damage to white matter tracts, it does pose a practical challenge to tractography algorithms, which rely on the aforementioned amplitude threshold to determine where and how far tractography may proceed.…”

“…Starting from the preprocessed pre-operative dMRI data, a novel technique to model the diffusion signal was applied, named single-shell 3-tissue constrained spherical deconvolution (SS3T-CSD) Dhollander, Raffelt, & Connelly, 2016), using MRtrix3Tissue (https://3Tissue.github.io), a fork of the MRtrix3 software (Tournier et al, 2019). Investigating the performance of this technique for this purpose, we have recently shown that SS3T-CSD allows to reconstruct white matter streamlines within infiltrative tumors and in immediately adjacent tissue (Aerts, Dhollander, & Marinazzo, 2019).…”

“…While this likely reflects the smaller portion of space taken up by axons (due to infiltrating tumor tissue) and/or damage to white matter tracts, it does pose a practical challenge to tractography algorithms, which rely on the aforementioned amplitude threshold to determine where and how far tractography may proceed. As detailed in Aerts et al (2019), we overcame this by gradually reducing the FOD amplitude threshold close to and even more so within the tumor, based on its prior segmentation. Anatomical constraints were also imposed to the generation of streamlines, informed by a segmented tissue image (Smith, Tournier, Calamante, & Connelly, 2012).…”