Machine-learning based guided diagnosis of parotid tumours from MRI
Résumé
Diagnosis of parotid gland tumours rely on magnetic resonance examination, fine-needle aspiration biopsy, or in some cases, invasive surgery that can damage facial nerves. In this work, we propose a machine learning model that can discriminate parotid tumours into histopathological subtypes from magnetic resonance imaging (MRI) scans, and further evaluate its impact on the diagnostic decisions of radiologists. We aim at improving the diagnosis of parotid neoplasms while avoiding any physical harm to patients. The radiologists improved their performance after observing the predictions of the algorithm. We conclude that machine-learning-based radiomics classification can assist radiologists.
Domaines
Intelligence artificielle [cs.AI]| Origine | Fichiers produits par l'(les) auteur(s) |
|---|
