A 3T Magnetic Resonance Fractional Order Calculus Model for the Diagnosis and Grading of Endometrial Cancer

To investigate MRI fractional order calculus model (FROC) on distinguishing different pathological grades of endometrial carcinoma (EC).

Fifty patients with pathologically-confirmed EC and normal volunteers were enrolled for analysis of parameters β, μ, and D.

The parameters β, μ, and D of the EC group were all significantly (P<0.05) lower than the normal controls. Parameters β, μ, and D had high diagnostic efficacy in distinguishing EC, with the AUC of β, μ, and D being 0.777, 0.660, and 0.809, respectively. The β Energy and Total Energy were significantly (P<0.05) different between G1 and G3 pathological grades of EC. The β Energy, Kurtosis, and Total Energy were significantly (P<0.05) different between G2 and G3 pathological grades and had AUCs of 0.730, 0.795, and 0.812, respectively. The D Entropy, Total Energy, and Energy showed a significant (P<0.05) difference between G2 and G3 grades and had AUCs of 0.724, 0.761, and 0.756, respectively. The μ Energy, Entropy, and Total Energy showed a significant (P<0.05) difference between G2 and G3 grades.

The use of parameters of the FROC model is effective for the diagnosis of endometrial carcinoma. Nonetheless, the limitations should be properly addressed, with validation of the model in a larger, multi-center cohort, incorporating second-order and high-order texture features.

The use of parameters of the FROC model is effective for the diagnosis of EC, and the first-order statistical parameters have high clinical value in differentiating high- and low-grade EC.