Methodology guidance for whole-brain connectivity in glioma

[Shasun K]

Dear Dr. Frank,

I am currently using DSI Studio to explore whole-brain connectivity in patients with motor eloquent gliomas, and I would greatly appreciate your guidance on the optimal tracking methodology for my scenarios.

Acquisition and preprocessing summary:

• Multishell diffusion: b = 1000 and 2500 s/mm², 64 directions each

• In-plane resolution 1.719 mm, slice thickness 2.60 mm, resampled to 2.00 mm isotropic

• Susceptibility distortion corrected with FSL topup (4 PA + 8 AP b0)

• Eddy current and motion correction with FSL eddy (132 DWI + 8 opposite phase DWI, b-table rotated)

• Bias field corrected using b0

• Reconstruction: generalized q-sampling imaging (GQI), diffusion sampling length ratio 1.25

• Tensor metrics calculated from b < 1750 s/mm²

My two possible tractography protocols are:

Protocol A — Augmented tracking strategies:

• Anisotropy threshold randomly selected between 0.5 and 0.7 Otsu

• Angular threshold randomly selected 45°–90°

• Step size = voxel spacing

• Track length 30–200 mm

• Tract-to-voxel ratio 1.0

Protocol B — Fixed thresholds:

• Anisotropy threshold 0.1 (or 0.05)

• Angular threshold 60°

• Step size 1.00 mm

• Track length 30–300 mm

• 1,000,000 seeds

My goal is whole-brain connectome analysis in these glioma patients, with particular interest in connectivity around the motor eloquent region, where peritumoral edema may degrade anisotropy and displace tracts.

My questions are:

1. For whole-brain connectivity analysis in motor eloquent glioma patients, would you recommend augmented tracking strategies (Protocol A) or fixed thresholds (Protocol B)?

2. If fixed thresholds are preferred, is an anisotropy threshold of 0.1 or 0.05 more appropriate for capturing tracts through peritumoral edema without introducing excessive false positives into the connectome?

3. Given the edema-related signal degradation in gliomas, would you advise switching from deterministic to probabilistic tracking for this patient population, or does the GQI model with augmented strategies sufficiently compensate?

4. Any additional parameter adjustments (e.g., diffusion sampling length ratio, angular threshold, step size) you would recommend specifically for tumor cases?

Please kindly provide your guidance in this scenario.

Thank you very much for your time and consideration.

Regards

Shanmugasundaram K