Pre-treatment MRI radiomics stratifies recurrence and survival risk in oral tongue cancer

In a 219-patient retrospective study of oral tongue squamous cell carcinoma, pre-treatment MRI radiomics combined with semantic, clinical, and pathology features predicted recurrence-free and overall survival, outperforming DeepSurv and reclassifying NCCN risk groups.

• The combined radiomics model (T1 + T2 + T1c radiomics + radiology semantic + clinical + pathology) achieved the best recurrence-free survival prediction and was non-inferior to the pathology model for overall survival prediction.
• Lasso-Cox outperformed DeepSurv for survival prediction; T1c radiomics best predicted metastatic lymph nodes, while radiology semantic features best predicted perineural invasion and pathologic extranodal extension.
• Pre-treatment radiomics risk groups reclassified NCCN risk categories, identifying low-risk patients within NCCN high-risk and high-risk within NCCN low + intermediate for recurrence; authors caution prospective multicenter validation is needed.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.