An individualized tractography pipeline for the nucleus basalis of Meynert lateral tract

At the center of the cortical cholinergic network, the nucleus basalis of Meynert (NBM) is crucial for the cognitive domains most vulnerable in Parkinson's disease (PD). Preclinical evidence has demonstrated the positive impact of NBM deep brain stimulation (DBS) on cognition but early human trials have had mixed results. It is possible that DBS of the lateral NBM efferent white matter fiber bundle may be more effective at improving cognitive-motor function. However, precise tractography modelling is required to identify the optimal target for neurosurgical planning. Individualized tractography approaches have been shown to be highly effective for accurately identifying DBS targets but have yet to be developed for the NBM.

Using structural and diffusion-weighted imaging, we developed a tractography pipeline using manually segmented regions of interest for precise individualized identification of the lateral NBM target tract. Using dice similarity coefficients, the reliability of the tractography outputs was assessed across three cohorts to investigate: 1) whether this manual segmentation pipeline is more reliable than an existing automatic segmentation pipeline currently used in the literature; 2) the inter- and intra-rater reliability of our pipeline in research scans of patients with PD; and 3) the reliability and practicality of this pipeline in clinical scans of DBS patients.

The individualized manual pipeline was found to be significantly more reliable than the existing automated pipeline for both the segmentation of the NBM region itself (p < 0.001) and the reconstruction of the target lateral tract (p = 0.002). There was also no significant difference between the reliability of two different raters in the PD cohort (p = 0.25), which showed high inter- (mean Dice coefficient >0.6) and intra-rater (mean Dice coefficient >0.7) reliability across runs. Finally, the pipeline was shown to be highly reliable within the clinical scans (mean Dice coefficient = 0.77). However, accurate reconstruction was only evident in 7/10 tracts.

We have developed a reliable tractography pipeline for the identification and analysis of the NBM lateral tract in research and clinical-grade imaging of healthy young adult and PD patient scans.