Radiomic Nomogram Improves Preoperative T Category Accuracy in Locally Advanced Laryngeal Carcinoma

摘要: Surgical decision-making on advanced laryngeal carcinoma is heavily depended the identification of preoperative T category (T3 vs. T4), which challenging for surgeons. A prediction radiomics (TCPR) model would be helpful subsequent surgery. total 211 patients with locally cancer who had undergone laryngectomy were randomly classified into training cohort (n = 150) and validation 61). We extracted 1,390 radiomic features from contrast-enhanced computed tomography images. Interclass correlation coefficient least absolute shrinkage selection operator (LASSO) analyses performed to select associated pathology-confirmed category. Eight found The signature was constructed by Support Vector Machine algorithm features. developed a nomogram incorporating reported experienced radiologists. performance evaluated area under curve (AUC). radiologists achieved an AUC 0.775 (95% CI: 0.667-0.883); while yielded significantly higher 0.862 0.772-0.952). predictive further improved, 0.892 0.811-0.974). Consequently, cancer, TCPR have great potential applied individual accurate may benefit regarding or larynx-preserving treatment.