Portrait de Kalum Ost n'est pas disponible

Kalum Ost

Collaborateur·rice de recherche
Superviseur⋅e principal⋅e
Sujets de recherche
Apprentissage automatique appliqué
Apprentissage automatique médical
IA en santé
Imagerie médicale
Neurosciences computationnelles

Publications

Advanced MRI scan acquisition metrics improve baseline disease severity predictions compared to traditional community MRI scan metrics
Abdul Al-Shawwa
David W. Cadotte
David Anderson
Nathan Evaniew
Nathan Evaniew
Bradley Jacobs
Julien Cohen‐Adad
Degenerative Cervical Myelopathy (DCM) is the functional derangement of the spinal cord and acts as one of the most common atraumatic spinal… (voir plus) cord injuries. Magnetic resonance imaging (MRI) are key in confirming the diagnosis of DCM in patients, though the utilization of higher fidelity magnetic resonance imaging scans and their integration into machine learning models remains largely unexplored. This study looks at the predictive ability of common community MRI scans in comparison to high fidelity scans in disease diagnosis. We hypothesize that the utilization of higher fidelity "advanced" MRI scans will increase the effectiveness of machine learning models predicting DCM severity. Through the utilization of Random Forest Classifiers, we have been able to predict disease severity with 41.8% accuracy in current community MRI scans and 63.9% in the advanced MRI scans. Furthermore, across the different predictive model variations tested, the advanced MRI scans consistently produced higher prediction accuracies compared to the community MRI counterparts. These results support our hypothesis and indicate that machine learning models have the potential to predict disease severity. However, neither performed well enough to be considered for use in clinical practice, indicating that the utilization of more sophisticated machine models may be required for these purposes.