Periodic Reporting for period 1 - SCIseg (Traumatic Spinal Cord Injury: The Need to Classify Disease Severity)
Reporting period: 2023-05-01 to 2025-04-30
This project aims to develop an automatic and reproducible analysis tool for magnetic resonance imaging (MRI) images based on machine learning and deep learning techniques, with the goal of extracting clinically relevant metrics to improve the management of patients with traumatic spinal cord injury (SCI). Each year, traumatic SCI affects between 250,000 and 500,000 individuals worldwide, often resulting from motor vehicle accidents, falls, or sports-related injuries. Traumatic SCI frequently causes severe neurological impairments, leading to a significant reduction in patients’ quality of life and imposing a substantial economic burden on healthcare systems. Although MRI examination is routinely performed in patients with traumatic SCI, its full potential remains underutilized due to the complexity of image analysis and the variability of MRI data across institutions. Deep learning, a field of artificial intelligence (AI), offers promising solutions by enabling automatic annotation, known as segmentation, of structures such as the spinal cord or lesions. This helps reduce inter-rater variability and supports the analysis of large, multi-center SCI cohorts. Quantitative MRI biomarkers derived using deep learning-based methods have already demonstrated strong correlations with clinical measures. Despite these advantages, deep learning applications in the context of traumatic SCI remain underexplored, and no open-source tools are currently available. To address this gap, the project focuses on developing an automatic, reproducible pipeline for MRI processing and metric extraction to support clinical decision-making in traumatic SCI.
I developed the first open-source deep learning model for automatic segmentation of the spinal cord and intramedullary lesions in both traumatic and non-traumatic SCI using T2-weighted MRI scans, trained on a diverse, multisite real-world dataset and integrated into an open-source software for spinal cord MRI data analysis. Building on this segmentation model, I created a comprehensive pipeline for computing quantitative MRI metrics of SCI severity and developed normalization methods to account for biological and scanner-related variability. These quantitative MRI-derived metrics, together with clinical scores and demographics, were used to develop a machine learning model predicting clinical outcomes. All outputs, including the deep learning models, code, and processing scripts, were released as open-source to ensure accessibility for the broader research community. The results have been presented at international and national conferences and published as open-access articles in high-impact peer-reviewed journals. Further communication and exploitation activities included presentations at clinical and scientific seminars, workshop organization, participation in competitions, social media outreach, and other scientific engagements, such as joining professional societies. More information is available on the public project website: https://janvalosek.com/pages/sciseg.html(opens in new window).
I developed a deep learning-based model for automatic segmentation of the spinal cord and intramedullary lesions in patients with traumatic and non-traumatic SCI using T2-weighted MRI scans. The model was developed and evaluated on a diverse, multisite real-world dataset and made publicly available as part of the open-source library for spinal cord MRI data analysis called Spinal Cord Toolbox. Notably, the model represents the first publicly available (open-source) tool for intramedullary lesion segmentation in spinal cord injury patients. As it segments jointly both the spinal cord and the lesions (instead of having two individual models), it goes beyond the current state of the art. The accompanying publication was published as an open-access peer-reviewed paper in the top-tier journal Radiology: Artificial Intelligence (Impact Factor: 8.1 Rank—Radiology, Nuclear Medicine & Medical Imaging: 9/204) and presented at an international conference in Singapore. Then, I updated the model (retrained it with additional new data we gained) and presented it at a workshop focused on applications of medical AI and published it as conference proceedings. The tool gained traction from researchers and clinicians around the world across disciplines. As the model can be applied to both traumatic and non-traumatic SCI patients, it goes beyond the original proposal (initially meant to be designed solely for traumatic SCI) and beyond the current state of the art.
Then, I employed the segmentation tool within a comprehensive analysis pipeline to compute quantitative MRI-derived metrics of the traumatic SCI severity on a large multi-site dataset. As the quantitative MRI-derived metrics might be impacted by MRI image parameters and biological factors (e.g. sex and age), I also focused on developing tools to mitigate these potential biases. Namely, I built a normative database of healthy participants and developed a normalization method allowing to normalize patients’ images relative to this database to account for biological factors such as sex and age and MRI scanner variability. Additionally, we developed an automatic method for quantitative metrics self-normalization to account for differences along the spinal cord anatomy. Finally, I automated the computation of several quantitative metrics, which have been so far computed only manually (which is a subjective, time-consuming task prone to intra- and inter-expert variability), thus going beyond the current state of the art.
After obtaining quantitative MRI-derived metrics using the developed tools, I used them, along with clinical scores and demographic data, to train a machine learning model for predicting patient outcomes. Several clinical outcomes were used as prediction targets, and a range of machine learning algorithms, optimization strategies, feature reduction techniques, and different feature sets were tested to evaluate their ability to predict patients’ clinical outcomes.
Then, I employed the segmentation tool within a comprehensive analysis pipeline to compute quantitative MRI-derived metrics of the traumatic SCI severity on a large multi-site dataset. As the quantitative MRI-derived metrics might be impacted by MRI image parameters and biological factors (e.g. sex and age), I also focused on developing tools to mitigate these potential biases. Namely, I built a normative database of healthy participants and developed a normalization method allowing to normalize patients’ images relative to this database to account for biological factors such as sex and age and MRI scanner variability. Additionally, we developed an automatic method for quantitative metrics self-normalization to account for differences along the spinal cord anatomy. Finally, I automated the computation of several quantitative metrics, which have been so far computed only manually (which is a subjective, time-consuming task prone to intra- and inter-expert variability), thus going beyond the current state of the art.
After obtaining quantitative MRI-derived metrics using the developed tools, I used them, along with clinical scores and demographic data, to train a machine learning model for predicting patient outcomes. Several clinical outcomes were used as prediction targets, and a range of machine learning algorithms, optimization strategies, feature reduction techniques, and different feature sets were tested to evaluate their ability to predict patients’ clinical outcomes.