Diffusion-Informed Joint Segmentation Enhances Detection of Thalamic Atrophy in Parkinson's Disease

Area of Science:

• Neuroimaging
• Neuroscience
• Medical Imaging

Background:

• The thalamus is a crucial subcortical structure involved in sensorimotor and cognitive functions.
• Accurate segmentation of thalamic nuclei is vital for understanding neurological diseases like Parkinson's disease (PD).
• Existing segmentation methods may not fully capture subtle volumetric changes associated with cognitive impairment.

Purpose of the Study:

• To compare the efficacy of a conventional structural MRI segmentation method with a joint framework integrating diffusion tensor imaging (DTI) for thalamic nuclei segmentation.
• To assess the sensitivity of each method in detecting volumetric differences in the thalamus of patients with Parkinson's disease and mild cognitive impairment (PD-MCI).
• To explore the association between thalamic subnucleus volumes and cognitive scores in PD-MCI patients.

Main Methods:

• Magnetic resonance imaging (MRI) data from healthy controls (HC), cognitively normal Parkinson's disease (PD-CN), and PD-MCI patients were analyzed.
• Two segmentation approaches were compared: conventional structural and a joint framework integrating DTI.
• Group comparisons evaluated volumetric differences and sensitivity to disease-related changes, with associations to cognitive scores (ACE-R) examined.

Main Results:

• Joint segmentation produced systematically lower thalamic volume estimates compared to the structural method.
• Joint segmentation, unlike structural segmentation, identified significant right thalamic atrophy in PD-MCI patients.
• Joint segmentation revealed reduced volumes in bilateral anterolateral and posterior thalamic nuclei in PD-MCI patients, which correlated positively with cognitive scores.

Conclusions:

• Methodological choices in segmentation significantly impact the detection of thalamic pathology in Parkinson's disease.
• Incorporating diffusion MRI into segmentation enhances sensitivity to subnuclear volume changes relevant to cognitive function in PD.
• This advanced segmentation approach holds promise for early diagnosis and monitoring of cognitive changes in Parkinson's disease.