MRI radiomics study strengthens case for canine glioma model
Bottom line
Version 1
A new comparative imaging study in Frontiers in Veterinary Science reports that radiomics extracted from T2-FLAIR MRI scans could distinguish tumoral from non-lesioned brain tissue across canine gliomas and human glioblastomas with 80.33% accuracy, adding support for dogs as a translational model in brain cancer research. The paper, accepted August 12, 2026, was authored by Ricardo Jorge Ferreira Faustino and Joaquim Pereira Henriques and appears in the journal’s oncology section and a comparative oncology research topic. Related thesis material describing the same work says the analysis identified radiomic features with area-under-the-curve values above 80%, including texture-based measures such as joint average and autocorrelation, and also found significant volumetric differences between the canine and human sample groups. (frontiersin.org)
Why it matters: For veterinary professionals, the study adds to a growing body of evidence that MRI-based radiomics may help move canine neuro-oncology beyond visual interpretation alone and toward more quantitative, noninvasive biomarker development. That could eventually support better case stratification, prognosis, and trial design in dogs with glioma, while also strengthening the role of naturally occurring canine disease in comparative oncology. Still, recent expert review literature emphasizes that most veterinary neuroimaging AI tools remain at the proof-of-concept stage and need standardized imaging protocols, reproducible pipelines, and external validation before routine clinical use. (onlinelibrary.wiley.com)
What to watch: The next step is whether this radiomics approach is validated in larger, multi-center canine cohorts and linked to clinically useful endpoints such as tumor subtype, treatment response, or survival. (research.ulusofona.pt)
Key facts
- Study type
- Comparative imaging study
- Journal
- Frontiers in Veterinary Science
- Imaging method
- T2-FLAIR MRI
- Species studied
- Canine gliomas and human glioblastomas
- Main finding
- Radiomics distinguished tumoral from non-lesioned brain tissue with 80.33% accuracy
- Radiomic features
- Joint average and autocorrelation were among the stronger discriminators
- Additional result
- Several radiomic markers had AUC values above 80%
- Other finding
- Significant differences in volume distribution were reported between the canine and human groups
- Acceptance date
- August 12, 2026
Version 2
A newly accepted study in Frontiers in Veterinary Science suggests that radiomics from routine T2-FLAIR MRI may help bridge canine and human brain tumor research. In “Radiomics and Comparative Neuroimaging of Canine Gliomas and Human Glioblastomas: A Step Towards Precision Medicine,” researchers reported 80.33% accuracy in distinguishing tumoral from non-lesioned brain tissue, using imaging-derived features from canine gliomas and human glioblastomas. The work adds fresh support to the idea that pet dogs with naturally occurring gliomas can serve as a clinically relevant comparative model for human glioblastoma research. (frontiersin.org)
That idea has been building for years. Comparative oncology groups, including the National Cancer Institute’s Comparative Brain Tumor Consortium, have argued that canine gliomas share meaningful histologic, biologic, and imaging similarities with human gliomas, and prior efforts have focused on standardizing classification to make canine cases more useful in translational research. Earlier veterinary imaging studies have also shown that machine learning and MRI texture analysis can classify canine brain tumors and, in some settings, predict glioma subtype or grade with promising accuracy. (pmc.ncbi.nlm.nih.gov)
According to the available study summary and related thesis record, the new analysis used segmented T2-FLAIR MRI regions of interest from canine gliomas and human glioblastomas to extract texture, shape, and intensity features. The authors reported that several radiomic markers had AUC values above 80%, with joint average and autocorrelation highlighted among the stronger discriminators. The work also found significant differences in volume distribution between the canine and human groups, with statistical significance reported at p < 0.001. The thesis record notes that the study was completed in 2025, while the journal listing shows the manuscript was accepted on August 12, 2026. (research.ulusofona.pt)
There does not appear to be a separate institutional press release or broad industry reaction available yet in the sources surfaced by web search. But the broader expert backdrop is clear: a 2026 review in Veterinary Medicine and Science says veterinary neuroimaging is moving from qualitative MRI interpretation toward quantitative AI-supported analysis, including radiomics, while cautioning that most tools are still research-stage. That review specifically points to the need for standardized acquisition, external validation, uncertainty reporting, and veterinarian-in-the-loop use before these methods are ready for clinical deployment. (onlinelibrary.wiley.com)
Why it matters: For veterinarians, especially neurologists, oncologists, and academic clinicians, this is less about an immediately practice-ready diagnostic tool and more about infrastructure for the next phase of canine brain tumor care. If radiomic signatures can be reproduced across scanners, institutions, and case mixes, they could support more objective diagnosis, biologic stratification, prognosis, and enrollment into clinical trials. That would be valuable in a disease area where biopsy is not always pursued and MRI interpretation can be challenging, particularly when trying to distinguish tumor biology or compare outcomes across studies. (onlinelibrary.wiley.com)
The comparative angle also matters. Naturally occurring canine gliomas may offer a more clinically realistic translational platform than induced laboratory models because they arise spontaneously, progress in an intact immune system, and are imaged and treated in real-world care settings. That’s one reason comparative oncology researchers have continued to invest in harmonizing pathology and imaging frameworks between species. This new study fits squarely within that trajectory, even if its current evidence base remains preliminary. (journals.sagepub.com)
What to watch: The key next milestone will be prospective or multi-center validation that ties these MRI features to outcomes veterinarians can act on, such as histologic subtype, therapeutic response, progression patterns, or survival, rather than tissue discrimination alone. (research.ulusofona.pt)