Abstract #2287

Can Hyperpolarized Gas MRI and Machine Learning Predict Longitudinal Changes in Airflow Limitation in Ex-smokers?

Maksym Sharma¹, Andrew R. Westcott¹, Aaron Fenster¹, David G. McCormack², and Grace Parraga³

¹Department of Medical Biophysics, Western University, London, ON, Canada, ²Department of Medicine, Western University, London, ON, Canada, ³Department of Medical Biophysics, Department of Medicine, Western University, London, ON, Canada

Hyperpolarized-gas-MRI provides a way to measure lung ventilation in patients with chronic obstructive pulmonary disease (COPD) in whom progressive worsening of expiratory airflow occurs over time. Progression of COPD is believed to stem from airway wall and lumen changes, airway remodeling or obliteration and emphysema. Our objective was to test machine-learning algorithms trained on hyperpolarized ³He MRI for predicting clinically-relevant FEV₁ changes. Novel 3-dimensional adaptations of gray-level run-length-matrices, gap-length-matrices, zone-size-matrices and co-occurrence-matrices were used for feature extraction which lead to the identification of features that predicted changes in airflow limitation (∆FEV₁%_pred>5%) over a 2.5-year time-period in at-risk and COPD ex-smokers.

This abstract and the presentation materials are available to members only; a login is required.

Join Here