CT Radiomics for Histological Grade Assessment of Canine
Soft Tissue Sarcoma: A Pilot Study
Seul Lee1, Jongin Kim2, Sungsoo
Kim3, Kichang Lee1,4
and Hakyoung Yoon1,4,5*
1Department
of Veterinary Medical Imaging, College of Veterinary
Medicine, Jeonbuk National University, Iksan, Jeonbuk-do,
54596, Republic of Korea; 2VIP Animal Medical
Center Cheongdam, 7, Samseong-ro 133-gil, Gangnam-gu,
Seoul, Republic of Korea; 3VIP Animal Medical
Center Seongbuk, Dongsomun-ro, Seongbuk-gu, Seoul, South
Korea;
4Vetradilab
Co., Ltd, Iksan 54596, Jeollabuk-do, Republic of Korea;
5Biosafety
Research Institute and College of Veterinary Medicine,
Jeonbuk National University, Iksan, Jeonbuk-do, 54596,
Republic of Korea
Histological grade is an important prognostic factor in
dogs with soft tissue sarcoma (STS) and is a key
consideration in surgical planning. Complete surgical
excision is the mainstay of treatment, and high-grade
tumors generally require wider surgical margins and
consideration of adjuvant therapy because of their
higher risk of recurrence and metastasis. Accurate
preoperative assessment of tumor grade is therefore
important for treatment planning, but preoperative
biopsy grading does not always match the final
histological grade after surgery. Therefore, additional
methods that can complement preoperative grade
assessment are needed. This exploratory study
investigated whether computed tomography (CT)-based
radiomic features differed between low- and high-grade
canine STS. Twenty-seven dogs with STS (20 low-grade [13
grade 1, 7 grade 2] and 7 high-grade [grade 3]) were
included in this study. From preoperative
contrast-enhanced CT images, 107 radiomic features were
extracted according to the Image Biomarker
Standardisation Initiative (IBSI) guidelines. Nine
features were then selected through a three-stage
process consisting of statistical filtering, exclusion
of volume-dependent features, and hierarchical
clustering. These features showed discriminatory
performance for differentiating low-grade from
high-grade tumors, with apparent AUC values of
0.843–0.957. GLDM-SmallDependenceLowGrayLevelEmphasis
had the highest apparent AUC of 0.957 (95% CI:
0.864–1.000). Low-grade tumors tended to show higher
values for low-gray-level texture features, whereas
high-grade tumors showed higher values for features
related to heterogeneous high-gray-level texture
patterns. These findings suggest that CT-based radiomics
may serve as a complementary tool for preoperative grade
assessment in canine STS. Although these findings are
encouraging, the reported AUC values reflect exploratory
performance within this study cohort, and future
prospective studies with larger cohorts and external
validation are needed to confirm these findings.
To Cite This Article:
Lee S, Kim J, Kim S, Lee K and Yoon H, 2026. CT
radiomics for histological grade assessment of canine
soft tissue sarcoma: a pilot study. Pak Vet J, 46(9):
2303-2310.
http://dx.doi.org/10.29261/pakvetj/2026.219