A novel synthetic MRI for discrimination of oral cavity and oropharynx malignancy

To evaluate the discriminative performance of morphologic and quantitative images derived from synthetic MRI in oral cavity and oropharynx lesions.

A total of 65 patients with 69 lesions and 5 healthy volunteers underwent preoperative MRI scans using the pre-contrast synthetic MRI sequence called MAGiC, which generates morphologic and quantitative images. The repeatability of the quantitative values was assessed in healthy volunteers using the coefficient of variation (CV) and Bland-Altman plots. The quantitative values were measured, and subsequent histogram analysis was conducted on the images obtained by MAGiC to differentiate between benign and malignant lesions. Logistic regression calculated prediction probabilities for each index, combining multiple indices to generate receiver operating characteristic (ROC) curves, which were compared using the Delong test.

The quantitative values derived from the MAGiC sequence exhibited high repeatability with CV below 15%. The directly measured T1 and T2 values within the lesions exhibited significant differences (P < 0.01). Moreover, notable differences were observed in the histogram features of Energy and Total Energy for differentiation (P < 0.05). The area under the curve (AUC) for histogram features extracted from synthetic morphologic images was 0.826, while for synthetic quantitative images was 0.833. The diagnostic accuracy was further enhanced by integrating synthetic morphologic and quantitative images, resulting in an optimal AUC of 0.936.

As a time-efficient synthetic sequence, MAGiC offers an alternative approach for preoperative differentiation of oral cavity and oropharynx lesions with enhanced diagnostic efficacy in the absence of contrast.

Not applicable.