# Deep Learning-Based Assessment of Brainstem Volume Changes in Spinocerebellar Ataxia Type 2 (SCA2): A Study on Patients and Preclinical Subjects

## TL;DR

A deep learning model accurately segments brainstem regions in SCA2 patients and preclinical subjects, revealing volume differences that could help track disease progression.

## Contribution

A modified U-Net with attention and inception modules achieves superior brainstem segmentation and reveals atrophy patterns in SCA2.

## Key findings

- The modified U-Net outperforms existing methods in brainstem segmentation with higher Dice scores across all substructures.
- Controls have significantly larger brainstem volumes compared to preclinical and patient groups, indicating atrophy linked to disease progression.
- Pons volume is reduced by nearly 50% in SCA2 patients and preclinical carriers compared to controls.

## Abstract

What are the main findings?
Superior Segmentation Performance: The proposed modified U-Net architecture (with attention-enhanced skip connections and inception modules) significantly outperforms three comparative approaches in brainstem parcellation, achieving higher scores across all substructures (medulla, pons, and mesencephalon) and the whole brainstem.Volume Differences Across Groups: Automated segmentation reveals distinct volumetric patterns, with controls exhibiting larger volumes (whole brainstem: 1.62) compared to preclinical (1.49) and patient groups (1.12), suggesting potential atrophy linked to disease progression.

Superior Segmentation Performance: The proposed modified U-Net architecture (with attention-enhanced skip connections and inception modules) significantly outperforms three comparative approaches in brainstem parcellation, achieving higher scores across all substructures (medulla, pons, and mesencephalon) and the whole brainstem.

Volume Differences Across Groups: Automated segmentation reveals distinct volumetric patterns, with controls exhibiting larger volumes (whole brainstem: 1.62) compared to preclinical (1.49) and patient groups (1.12), suggesting potential atrophy linked to disease progression.

What is the implication of the main finding?
Clinical Utility: The method’s accuracy and robustness support its potential for precise brainstem assessment in neurodegenerative disorders, enabling earlier detection of structural changes (e.g., reduced medulla volume in patients: 0.26 vs. 0.31 in controls).Technical Advancements: The success of attention mechanisms and inception modules highlights their value for complex anatomical segmentation, paving the way for similar adaptations in other small-structure parcellation tasks.

Clinical Utility: The method’s accuracy and robustness support its potential for precise brainstem assessment in neurodegenerative disorders, enabling earlier detection of structural changes (e.g., reduced medulla volume in patients: 0.26 vs. 0.31 in controls).

Technical Advancements: The success of attention mechanisms and inception modules highlights their value for complex anatomical segmentation, paving the way for similar adaptations in other small-structure parcellation tasks.

Spinocerebellar ataxia type 2 (SCA2) is a neurodegenerative disorder marked by progressive brainstem and cerebellar atrophy, leading to gait ataxia. Quantifying this atrophy in magnetic resonance imaging (MRI) is critical for tracking disease progression in both symptomatic patients and preclinical subjects. However, manual segmentation of brainstem subregions (mesencephalon, pons, and medulla) is time-consuming and prone to human error. This work presents an automated deep learning framework to assess brainstem atrophy in SCA2. Using T1-weighted MRI scans from patients, preclinical carriers, and healthy controls, a U-shaped convolutional neural network (CNN) was trained to segment brainstem subregions and quantify volume loss. The model achieved strong agreement with manual segmentations, significantly outperforming four U-Net-based benchmarks (mean Dice scores: whole brainstem 0.96 vs. 0.93–0.95, pons 0.96 vs. 0.91–0.94, mesencephalon 0.96 vs. 0.89–0.93, and medulla 0.95 vs. 0.91–0.93). Results revealed severe atrophy in preclinical and symptomatic cohorts, with pons volumes reduced by nearly 50% compared to controls (p < 0.001). The mesencephalon and medulla showed milder degeneration, underscoring regional vulnerability differences. This automated approach enables rapid, precise assessment of brainstem atrophy, advancing early diagnosis and monitoring in SCA2.