Abstract: 【Abstract】 Objective To investigate the clinical value of high-resolution magnetic resonance imaging （HRMRI） and high-frequency ultrasound in the preoperative evaluation of cutaneous melanoma. Methods This study was a diagnostic accuracy study. Clinical, imaging, and pathological data were collected from 18 patients with melanoma, who were treated at the First Affiliated Hospital of Soochow University between January 2019 and June 2022. All the 18 patients underwent HRMRI and histopathological examination, and 9 of them additionally underwent high-frequency ultrasound examination. Multimodal imaging features of melanoma were analyzed. By comparing with histopathological findings, the accuracy of HRMRI and high-frequency ultrasound was evaluated for distinguishing between melanotic melanoma and hypomelanotic melanoma, measuring tumor invasion depth, and assessing satellite lesions. The paired t-test, Wilcoxon signed-rank test, or paired chi-square test was used for comparisons between paired groups. The intraclass correlation coefficient （ICC） or Kappa analysis was employed for consistency assessment. Results Among the 18 patients with melanoma, 7 were males and 11 were females, with ages of 67 ± 12 years （range, 52 - 81 years）. HRMRI revealed that most of the lesions were irregular in shape, with unclear borders and slightly uneven bases; melanotic melanoma lesions were slightly hyperintense to hyperintense on T1-weighted images （T1WI）, and isointense to hypointense on T2-weighted images （T2WI）; hypomelanotic melanoma lesions were isointense to hypointense on T1WI, and isointense to slightly hyperintense on T2WI; both of the above melanoma subtypes presented with homogeneous or heterogeneous signal intensity, and showed moderate to marked enhancement that was either homogeneous or heterogeneous on contrast-enhanced scans. High-frequency ultrasound revealed that all melanoma lesions were hypoechoic with posterior acoustic shadowing, and showed heterogeneous hyperenhancement on contrast-enhanced ultrasound images. For the differential diagnosis of hypomelanotic melanoma, there was no significant difference between the performance of HRMRI and that of histopathological findings （paired chi-square test, P = 1.000）, indicating good agreement （Kappa = 0.67, P = 0.004）. In contrast, high-frequency ultrasound showed that both melanoma subtypes were hypoechoic, and could not differentiate between them. For the evaluation of tumor invasion depth, no significant differences were observed between HRMRI or high-frequency ultrasound and histopathological findings （both P > 0.05）, and the ICCs with 95% confidence intervals （CIs） were 0.803 （0.471, 0.928） for HRMRI and 0.738 （0.002, 0.939） for high-frequency ultrasound, indicating good and moderate agreement, respectively. For the assessment of satellite lesions, HRMRI demonstrated a sensitivity of 100% （2/2; 95% CI: 20%, 100%） and a specificity of 69% （11/16; 95% CI: 41%, 88%）, while high-frequency ultrasound had a sensitivity of 0 （0/1; 95% CI: 0, 94%） and a specificity of 100% （8/8; 95% CI: 60%, 100%）. Conclusions HRMRI demonstrated good performance in diagnosing hypomelanotic melanoma and in assessing tumor invasion depth, with a strong capability for identifying satellite lesions. High-frequency ultrasound showed good performance in ruling out satellite lesions. Both modalities may have clinical value in the preoperative evaluation of cutaneous melanoma.

Key words: Melanoma, Skin, High-resolution magnetic resonance imaging, High-frequency ultrasound, Histopathological manifestations