Machine learning predicts histologic type and grade of canine gliomas based on MRI texture analysis

Barge P, Oevermann A, Maiolini A, Durand A. Machine learning predicts histologic type and grade of canine gliomas based on MRI texture analysis. Vet Radiol Ultrasound. Published online May 3, 2023. doi:10.1111/vru.13242

This study investigated the accuracy of machine learning (ML)-based MRI texture analysis (MRI-TA) for predicting the histologic types and grades of canine gliomas. Thirty-eight dogs with 40 intracranial gliomas were included, with tumors manually segmented across various MRI sequences. The ML models demonstrated an overall accuracy of 77% for tumor type classification and 75.6% for predicting high-grade gliomas. The support vector machine classifier achieved the highest accuracy, with 94% for tumor types and 87% for high-grade gliomas, with key predictive features found in peri-tumoral edema and non-enhancing tumor areas.