Volumetric analysis and MRI radiomics for response assessment in canine glioma immunotherapy: an exploratory feasibility study

Canine gliomas (CG) are a group of aggressive neurological tumours representing 2–5% of canine cancers. Although relatively uncommon compared with other neoplasms, CG are clinically significant due to poor prognosis and limited treatment options, markedly affecting quality of life and survival. In this work, we perform a longitudinal evaluation of tumour volumetric changes over time and investigate whether MRI-derived radiomic features can be used to develop exploratory predictive models of response to the oncolytic virus ICOCAV15. Volumetric analyses revealed early post-treatment reductions in responders, detectable within 15 days. The best-performing classification model, which used pre-treatment T2-weighted and FLAIR images, achieved 88% accuracy, 92% sensitivity, and 79% specificity in this exploratory cohort, suggesting that MRI-derived radiomic features may contain information associated with treatment response. The most informative radiomic features were derived from T2-weighted and FLAIR sequences and primarily captured characteristics related to tumour burden and morphology. These findings suggest the feasibility of integrating MRI, radiomics, and machine learning for the exploratory assessment of treatment response in canine gliomas. The identified radiomic features should be considered candidate imaging biomarkers that warrant further validation in larger prospective studies. While the present results are preliminary, they highlight the potential of quantitative imaging to support standardized response assessment and reinforce the value of canine gliomas as a translational model for future neuro-oncology and immunotherapy research.