Quantitative Susceptibility Atlas Construction in Montreal Neurological Institute Space: Towards Histological-consistent Iron-rich Deep Brain Nucleus Sub-region Identification

Chenyu He, Xiaojun Guan, Weimin Zhang, Jun Li, Chunlei Liu, Hongjiang Wei, Xiaojun Xu, Yuyao ZhangResearch 2021 · DOI

Abstract

Iron-rich deep brain nuclei (DBN) are vital for motor, emotional, and cognitive functions, with abnormal iron levels linked to various neurological diseases. While Quantitative Susceptibility Mapping (QSM) excels at imaging these structures, existing atlases often lack the resolution or standard alignment necessary for precise research. This study introduces the MuSus-100 atlas, a Montreal Neurological Institute (MNI) space unbiased QSM human brain atlas constructed from 100 healthy subjects. Guided by multi-modal fused "hybrid" images, the atlas provides exceptional contrast for DBN. Notably, it includes a histologically consistent parcellation of the thalamus into 64 sub-regions. Validation shows superior segmentation accuracy and the ability to identify pathology-related iron alterations in Parkinson’s Disease (PD) patients.

Introduction: The Challenge of Deep Brain Imaging

Deep brain nuclei (DBN), including the thalamus and basal ganglia, are critical nodes in neural circuits. Their high iron content makes them prime targets for Quantitative Susceptibility Mapping (QSM), which maps tissue magnetic susceptibility. However, standard structural MRI (T1w or T2w) often fails to provide sufficient contrast to distinguish small DBN sub-regions. Furthermore, a lack of high-resolution QSM templates in the standard MNI coordinate system has hindered large-scale group studies. The MuSus-100 atlas aims to bridge this gap by providing a high-quality, standardized susceptibility template.

Methods: Constructing the MuSus-100 Atlas

The researchers developed a sophisticated pipeline to ensure both cortical and deep-tissue alignment accuracy.

1. Hybrid Image Fusion and Registration

The process begins with the creation of "hybrid images" that combine T1-weighted data (for cortical boundaries) and QSM data (for deep nuclei). These hybrid images guide a group-wise registration process involving 100 healthy young adults (ages 19-29). This iterative registration ensures the final template is an unbiased population average.

Overview of the MuSus-100 atlas construction pipeline, including (a) atlas space construction and (b) MNI space projection.

2. Border Enhancement and Thalamic Parcellation

To improve manual labeling accuracy, a 3D volumetric Laplacian kernel was applied to sharpen the edges of the susceptibility maps. For the thalamus, the team utilized a "joint label fusion" strategy. They manually labeled 64 sub-regions on a high-resolution 7-Tesla MRI scan—referencing the Schaltenbrand and Wahren histology atlas—and then warped these labels onto the 3-Tesla MNI-space atlas.

Key Insights: A "Best of Both Worlds" Strategy

The defining feature of the MuSus-100 atlas is its multi-modal fusion approach. By leveraging the strengths of different MRI modalities, the researchers achieved superior alignment.

Detail of the Fusing Module, showing how T1w and QSM data are combined to create a hybrid image for registration.
  • Thalamic Granularity: Unlike previous atlases that treat the thalamus as a few large zones, MuSus-100 provides a 64-region parcellation, bringing microscopic histological detail into a standard clinical coordinate system.
  • Diagnostic Utility: The atlas was validated using a cohort of 87 subjects (47 PD patients and 40 controls). It successfully identified specific iron accumulation in the ventral and central regions of the thalamus in PD patients, highlighting its potential as a biomarker tool.

Results and Clinical Application

The MuSus-100 atlas demonstrated significant improvements over existing benchmarks (e.g., Colin-27, ICBM-152):

  • Segmentation Accuracy: Achieved Dice coefficients between 0.73 and 0.90, outperforming other atlases by 5-10%.
  • Pathology Identification: Pinpointed the centralis lateralis and paracentral nuclei as key areas of iron alteration in Parkinson's Disease.

Limitations and Future Directions

While a significant step forward, the atlas is currently based on a young adult population (average age ~24). Given that brain iron levels change with age, future iterations should include longitudinal data to create age-specific templates. Additionally, while 7T-guided, the final atlas resolution is 1mm isotropic; sub-millimetric population templates remain a goal for future research.

Summary updated

Abstract

Abstract Iron-rich deep brain nuclei (DBN) of the human brain are involved in various motoric, emotional and cognitive brain functions. The abnormal iron alterations in the DBN are closely associated with multiple neurological and psychiatric diseases. Quantitative susceptibility mapping (QSM) provides the spatial distribution of tissue magnetic susceptibility in the human brain. Compared to traditional structural imaging, QSM has superiority for imaging the iron-rich DBN owing to the susceptibility difference existing between brain tissues. In this study, we construct a Montreal Neurological Institute (MNI) space unbiased QSM human brain atlas via group-wise registration from 100 healthy subjects aged 19-29 years. The atlas construction process is guided by hybrid images that fused from multi-modal Magnetic Resonance Images (MRI), thus named as Multi-modal-fused magnetic Susceptibility (MuSus-100) atlas. The high-quality susceptibility atlas provides extraordinary image contrast between iron-rich DBN with their surroundings. Parcellation maps of DBN and their sub-regions that are highly related to neurological and psychiatric pathology are then manually labeled based on the atlas set with the assistance of an image border-enhancement process. Especially, the bilateral thalamus is delineated into 64 detailed sub-regions referring to the Schaltenbrand and Wahren stereotactic atlas. To our best knowledge, the histological-consistent thalamic nucleus parcellation map is well defined for the first time in MNI space. Comparing with existing atlases emphasized on DBN parcellation, the newly proposed atlas outperforms on atlas-guided individual brain image DBN segmentation accuracy and robustness. Moreover, we apply the proposed DBN parcellation map to conduct detailed identification of the pathology-related iron content alterations in subcortical nuclei for Parkinson Disease (PD) patients. We envision that the MuSus-100 atlas could play a crucial role in improving the accuracy of DBN segmentation for the research of neurological and psychiatric disease progress and also be helpful for target planning in deep brain stimulation surgery
BibTeX
@inproceedings{he2021quantitative,
  author = {Chenyu He and Xiaojun Guan and Weimin Zhang and Jun Li and Chunlei Liu and Hongjiang Wei and Xiaojun Xu and Yuyao Zhang},
  title = {{Quantitative Susceptibility Atlas Construction in Montreal Neurological Institute Space: Towards Histological-consistent Iron-rich Deep Brain Nucleus Sub-region Identification}},
  booktitle = {Research},
  year = {2021},
  doi = {10.21203/rs.3.rs-994193/v1},
}
Figure