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

Pablo Barge, Anna Oevermann, Arianna Maiolini, Alexane Durand

Background: The authors aimed to investigate the accuracy of machine learning (ML) models based on MRI texture analysis (TA) in predicting canine gliomas histologic types and grades, as conventional MRI features have low sensitivity and specificity for this purpose.

Study: This was a retrospective, diagnostic accuracy study that included 38 dogs with histopathological diagnosis of intracranial glioma and available brain MRI. The tumors were segmented into enhancing, non-enhancing, and edema regions, and texture features were extracted from each region in four MRI sequences. Three ML classifiers (support vector machine, random forest, and k-nearest neighbor) were trained and tested using different datasets and cross-validation techniques.

Methods: The performance of the ML classifiers was assessed using accuracy, sensitivity, specificity, and area under the curve (AUC) for multiclass (oligodendroglioma vs. astrocytoma vs. oligoastrocytoma) and binary (high-grade vs. low-grade) classification models. The most discriminative texture features for each classification task were also identified.

Results: The ML classifiers had an average accuracy of 77% for discriminating tumor types and 76% for predicting high-grade gliomas on the whole dataset. The support vector machine classifier had the best performance, with an accuracy of up to 94% for predicting tumor types and up to 87% for predicting high-grade gliomas3. The most discriminative texture features for differentiating tumor types and grades appeared related to the peri-tumoral edema in T1-weighted images and to the non-enhancing part of the tumor in T2-weighted images, respectively.

Limitations: The study had several limitations, such as the small and imbalanced sample size, the use of different slice thicknesses and coils, the manual segmentation of the tumors, the lack of differentiation between gliomas and benign mass lesions, and the unknown prognostic relevance of glioma types and grades in dogs.

Conclusions: The authors concluded that ML models based on MRI-TA have the potential to discriminate intracranial canine glioma types and grades with high accuracy, and that texture features reflect histopathological characteristics not captured by conventional MRI. They suggested further multicenter studies with larger sample sizes to corroborate these results.

Disclaimer: The summary generated in this email was created by an AI large language model. Therefore errors may occur. Reading the article is the best way to understand the scholarly work. The figure presented here remains the property of the publisher or author and subject to the applicable copyright agreement. It is reproduced here as an educational work. If you have any questions or concerns about the work presented here, reply to this email.