Longitudinal MRI in asymptomatic degenerative cervical cord compression

1843 

Abstract Submission 

Jan Valosek^1,2,3,4, Petr Hluštík^5,6, Tomáš Horák^7,8, Magda Horáková^7,8, Josef Bednařík^7,8

¹NeuroPoly Lab, Institute of Biomedical Engineering, Polytechnique Montreal, Montreal, Quebec, ²Mila - Quebec AI Institute, Montreal, Canada, ³Department of Neurosurgery, Faculty of Medicine and Dentistry, Palacký University Olomouc, Olomouc, Czech Republic, ⁴Department of Neurology, Faculty of Medicine and Dentistry, Palacký University Olomouc, Olomouc, Czech Republic, ⁵Department of Neurology, Faculty of Medicine and Dentistry, Palacký University, Olomouc, Czech Republic, ⁶Department of Neurology, University Hospital, Olomouc, Czech Republic, ⁷Faculty of Medicine, Masaryk University, Brno, Czech Republic, ⁸Department of Neurology, University Hospital Brno, Brno, Czech Republic

Jan Valosek  
NeuroPoly Lab, Institute of Biomedical Engineering, Polytechnique Montreal|Mila - Quebec AI Institute|Department of Neurosurgery, Faculty of Medicine and Dentistry, Palacký University Olomouc|Department of Neurology, Faculty of Medicine and Dentistry, Palacký University Olomouc
Montreal, Quebec|Montreal, Canada|Olomouc, Czech Republic|Olomouc, Czech Republic

Petr Hluštík  
Department of Neurology, Faculty of Medicine and Dentistry, Palacký University|Department of Neurology, University Hospital
Olomouc, Czech Republic|Olomouc, Czech Republic

Tomáš Horák  
Faculty of Medicine, Masaryk University|Department of Neurology, University Hospital Brno
Brno, Czech Republic|Brno, Czech Republic

Magda Horáková  
Faculty of Medicine, Masaryk University|Department of Neurology, University Hospital Brno
Brno, Czech Republic|Brno, Czech Republic

Josef Bednařík  
Faculty of Medicine, Masaryk University|Department of Neurology, University Hospital Brno
Brno, Czech Republic|Brno, Czech Republic

Spinal cord morphometric measures computed from magnetic resonance (MRI) images, such as cross-sectional area and anteroposterior diameter, are widely used to assess the severity of spinal cord compression (Badhiwala et al. 2020). The longitudinal monitoring of the compression severity might play an important role in the surgical management of participants with asymptomatic degenerative cervical cord compression, particularly prior to the progression to degenerative cervical myelopathy. However, due to difficulties in robustly co-registering spinal cord MRI images across sessions, the longitudinal MRI evaluation is usually performed on morphometric measures aggregated across multiple levels, such as vertebral levels C2-C3 (David et al. 2021). As compression typically occurs in proximity to the intervertebral discs spanning two adjacent vertebral levels, the averaging across levels can thus impact compression parameters. In this study, we perform a longitudinal assessment of compression severity using a recently proposed normalization approach bringing morphometric measures from different sessions into common anatomical dimensions (Valošek et al. 2023).

3T T2-weighted images covering the cervical spinal cord were acquired in 59 participants with asymptomatic degenerative cervical cord compression. Participants were scanned longitudinally with an interval of 36 months between measurements. Image processing was done with the Spinal Cord Toolbox (SCT, v6.1) (De Leener et al. 2017). For each participant, the spinal cord was segmented using a deep learning-based algorithm trained on spinal cord injury patients (https://github.com/ivadomed/model_seg_sci), the vertebral levels were identified (Ullmann et al. 2014), and morphometric measures were normalized to the PAM50 anatomical dimensions (Valošek et al. 2023). The following morphometric measures were used: cross-sectional area, anteroposterior diameter, transverse diameter, compression ratio, eccentricity, and solidity.

Figure 1 shows morphometric measures for three participants with asymptomatic degenerative cervical cord compression (A: female, 53 y.o., B: female, 58 y.o., C: male, 53 y.o.). For each participant, the morphometric measures from both sessions are shown in the same anatomical dimensions of the PAM50 spinal cord template. The morphometric measures are plotted across individual axial slices with vertebral levels identified on the plot. The figure also shows (in gray color) normative values computed from a database of 203 healthy controls (Valošek et al. 2023).
For the whole patient group, the evolution of morphometric parameters over the follow-up interval of approximately 3 years appears quite heterogeneous. Several scenarios may be observed: First, in ~37% of the patients, the morphometry remains stable, without signs of either worsening or improvement. This may confirm the test-retest stability of the methodology. Second, ~42% of the patients present with worsening cord compression, either more pronounced compression in one segment or a new compression in another segment). Finally, ~20% of patients' compression seems to improve, so that the morphometric measures return to the normal range.

In conclusion, this study employs a normalization approach to longitudinally assess compression severity in participants with cervical spinal cord compression. Bringing morphometric measures from different sessions into common anatomical dimensions facilitates an assessment of all individual axial slices without averaging across levels. Subsequent work will focus on the validation of inter-session variability and correlation with patients' clinical status progression.

Image Registration and Computational Anatomy ¹

Segmentation and Parcellation ²

Segmentation

Spatial Normalization

Spinal Cord

STRUCTURAL MRI