A deep learning-based radiomic nomogram derived from visceral fat for early prediction of gastrointestinal stromal tumor risk grade
By
Wei Chen
Long-Yu Duan
Kun-Ming Yi
Xiao-Juan Peng
Lian-Qin Kuang
June 19, 2026
Clinical Scorecard: A Radiomic Nomogram Utilizing Deep Learning from Visceral Fat for Early Assessment of Gastrointestinal Stromal Tumor Risk Levels
At a Glance
Category Detail
Condition Gastrointestinal Stromal Tumors (GISTs) Key Mechanisms Utilizes deep learning-based radiomics to assess visceral fat features from non-contrast CT for risk stratification. Target Population Patients with histologically confirmed GISTs. Care Setting Preoperative risk assessment in clinical oncology.
Key Highlights
Developed a deep-learning-based radiomics nomogram (DLRN) for GIST risk grading. Achieved an AUC of 0.936 in the derivation cohort and 0.862 in the external test cohort. Demonstrated higher net clinical benefit compared to traditional models. Integrates visceral fat-derived features for non-invasive risk stratification. Addresses limitations of contrast-enhanced imaging in certain patient populations. Guideline-Based Recommendations
Diagnosis
Use of NIH criteria and AFIP classification for GIST risk stratification. Management
High-risk patients should receive at least 3 years of imatinib adjuvant therapy. Intermediate-risk patients with gastric primary tumors should undergo at least 1 year of imatinib adjuvant therapy. Monitoring & Follow-up
Very low-risk and low-risk patients should undergo regular follow-up. Risks
Limitations of postoperative pathological indicators for risk assessment. Patient & Prescribing Data
Patients with gastrointestinal stromal tumors (GISTs) requiring preoperative risk assessment.
Visceral fat may provide additional metabolic and inflammatory information for evaluation.
Clinical Best Practices
Consider non-contrast CT for patients with contraindications to contrast media. Utilize deep learning techniques for improved feature extraction in radiomics. Related Resources & Content
