Generation and evaluation of an ultra-high-field atlas with applications in DBS planning

Abstract

Purpose Deep brain stimulation (DBS) is a common treatment for Parkinson's disease (PD) and involves the use of brain atlases or intrinsic landmarks to estimate the location of target deep brain structures, such as the subthalamic nucleus (STN) and the globus pallidus pars interna (GPi). However, these structures can be difficult to localize with conventional clinical magnetic resonance imaging (MRI), and thus targeting can be prone to error. Ultra-high-field imaging at 7T has the ability to clearly resolve these structures and thus atlases built with these data have the potential to improve targeting accuracy. Methods T1 and T2-weighted images of 12 healthy control subjects were acquired using a 7T MR scanner. These images were then used with groupwise registration to generate an unbiased average template with T1w and T2w contrast. Deep brain structures were manually labelled in each subject by two raters and rater reliability was assessed. We compared the use of this unbiased atlas with two other methods of atlas-based segmentation (single-template and multi-template) for subthalamic nucleus (STN) segmentation on 7T MRI data. We also applied this atlas to clinical DBS data acquired at 1.5T to evaluate its efficacy for DBS target localization as compared to using a standard atlas. Results The unbiased templates provide superb detail of subcortical structures. Through one-way ANOVA tests, the unbiased template is significantly (p < 0.05) more accurate than a single-template in atlas-based segmentation and DBS target localization tasks. Conclusion The generated unbiased averaged templates provide better visualization of deep brain nuclei and an increase in accuracy over single-template and lower field strength atlases.

Automatic localization of the subthalamic nucleus on patient‐specific clinical MRI by incorporating 7 T MRI and machine learning: Application in deep brain stimulation

“…For this reason, we compared the performance of the proposed approach with that of state-of-the-art atlas-based methods: (1) 7 T3 T (Milchenko et al, 2018), (2) UHFA (Wang et al, 2016), (3) DISTAL (Ewert et al, 2017;Horn, Neumann, et al, 2017), and (4) MNIPD25 (Xiao et al, 2017). While these atlases reasonably localize and visualize the STN on subject-specific data they represent well, their accuracy and consistency on our tested clinical data were much lower than the proposed 7 T-ML method.…”

“…More specifically, the DISTAL atlas (Ewert et al, 2017) and UHFA (Wang et al, 2016) were defined on the atlas template from normal subjects, and segmentation results on data from normal subjects were provided. The results on the PD-specific data might be deteriorated by morphological variability between the atlas template and the clinical data.…”

“…The results on the PD-specific data might be deteriorated by morphological variability between the atlas template and the clinical data. Moreover, UHFA (Wang et al, 2016) utilized the 7 T T 2 W MRI atlas template, and thus the contrast discrepancy between the 7 T MRI and standard clinical data also might have affected the registration (this might also explain the often smaller STN volume found when computed with this atlas compared with other atlases-based results).…”

“…Probabilistic atlas maps obtained from multiple 7 T MR contrasts were used for analyzing anatomical variability on subcortical structures (Keuken et al, 2014). Wang et al (2016) generated the ultrahigh-field atlas of the STN using the 7 T T 1 -weighted (T 1 W) and T 2 -weighted (T 2 W) MRI. For use with clinical data, Plassard et al (2017) created atlases based on the 7 T MRI of healthy subjects and used them to segment subcortical structures on 3 T contrast enhanced MR sequences.…”

A framework for evaluating correspondence between brain images using anatomical fiducials

“…Here, we use these terms to describe (virtual, image-based) anatomical fiducials (AFIDs) annotated in structural T1-weighted MRI scans. Phase 1: Protocol validation for brain templates Novice participants (N=8) were trained over a series of neuroanatomy tutorials to place AFIDs on a number of publicly available brain images: Agile12v2016 (Lau et al, 2017;Wang et al, 2016), Colin27 (Holmes et al, 1998), MNI2009bAsym (nonlinear asymmetric; version 2009b; RRID:SCR_008796) (Fonov et al, 2011). Each participant then performed 4 rating sessions independently for each template, for a total of 12 point sets resulting in a total of 96 protocols.…”