Diffusion Relaxation—Correlated Spectroscopic MR Imaging for In Vivo Tumor Heterogeneity and Lymph Node Metastasis Prediction in Oral Tongue Squamous Cell Carcinoma

ABSTRACT Background In oral tongue squamous cell carcinoma (OTSCC), cervical lymph node metastasis (CLNM) reduces predicted survival by 50%. However, MRI sensitivity for occult nodal disease is only 35.7%, reflecting limited microstructural specificity. Purpose To investigate whether diffusion‐relaxation correlated spectroscopic imaging (DR‐CSI) metrics with peritumoral extensions contribute toward the prediction of CLNM in OTSCC. Study Type Prospective. Population 99 patients (33 female and 66 male) with pathologically confirmed OTSCC. Field Strength/Sequence 3T; a spin‐echo echo‐planar DR‐CSI with 35 contrasts. Assessment Tumor‐stroma ratio (TSR), density of tumor‐infiltrating lymphocytes (TILs), perineural invasion (PNI) and differentiation status were determined from histology. Tumor greatest diameter and depth of invasion (DOI) were evaluated from MRI. Apparent diffusion coefficient (ADC) maps and T 2 maps were derived from DR‐CSI data. Five DR‐CSI compartmental metrics in intratumoral and peritumoral regions, V A – V E , were calculated. Statistical Tests Mann–Whitney U‐test, Chi‐square test, or Spearman correlation analysis. Multivariable logistic regression and receiver operating characteristic (ROC) analysis for diagnostic performance. Significance criteria: p < 0.05. Results 51 patients were CLNM negative. Intratumoral V A in the CLNM‐positive group (34.80% ± 18.51%) was significantly lower than that in the CLNM‐negative group (47.86% ± 23.37%), while V B (25.07% ± 12.49%) and V D (22.12% ± 11.46%) in the CLNM‐positive group were significantly higher than those in the CLNM‐negative group. ADC and T 2 showed no significant separation ( p = 0.084, 0.493, respectively). A multivariable logistic model comprising the significant intratumoral and peritumoral DR‐CSI metrics ( V A , V B , and V D ) together with the morphologic variables (tumor greatest diameter and DOI) achieved an area under curve (AUC) value of 0.792 for CLNM classification, representing a non‐significant improvement over the morphologic variables model alone (AUC 0.709, p = 0.069). Data Conclusion The intratumoral and peritumoral DR‐CSI metrics, combined with morphologic variables, improved preoperative discrimination of nodal status in OTSCC in this single center cohort, though external validation is warranted. Evidence Level 2. Technical Efficacy Stage 2.