Ultrasound Feature-Based Diagnostic Model Focusing on the “Submarine Sign” for Epidermal Cysts among Superficial Soft Tissue Lesions
- Author:
Da Hyun LEE
1
;
Choon Sik YOON
;
Beom Jin LIM
;
Hye Sun LEE
;
Sinae KIM
;
A Lam CHOI
;
Sungjun KIM
Author Information
- Publication Type:Original Article
- Keywords: Epidermal cyst; Submarine sign; Ultrasound; Nomogram
- MeSH: Acoustics; Calibration; Discrimination (Psychology); Epidermal Cyst; Humans; Logistic Models; Nomograms; Retrospective Studies; ROC Curve; Ultrasonography
- From:Korean Journal of Radiology 2019;20(10):1409-1421
- CountryRepublic of Korea
- Language:English
- Abstract: OBJECTIVE: To develop a diagnostic model for superficial soft tissue lesions to differentiate epidermal cyst (EC) from other lesions based on ultrasound (US) features. MATERIALS AND METHODS: This retrospective study included 205 patients who had undergone US examinations for superficial soft tissue lesions and subsequent surgical excision. The study population was divided into the derivation set (n = 112) and validation set (n = 93) according to the imaging date. The following US features were analyzed to determine those that could discriminate EC from other lesions: more-than-half-depth involvement of the dermal layer, “submarine sign” (focal projection of the hypoechoic portion to the epidermis), posterior acoustic enhancement, posterior wall enhancement, morphology, shape, echogenicity, vascularity, and perilesional fat change. Using multivariable logistic regression, a diagnostic model was constructed and visualized as a nomogram. The performance of the diagnostic model was assessed by calculating the area under the curve (AUC) of the receiver operating characteristic curve and calibration plot in both the derivation and validation sets. RESULTS: More-than-half-depth involvement of the dermal layer (odds ratio [OR] = 3.35; p = 0.051), “submarine sign” (OR = 12.2; p < 0.001), and morphology (OR = 5.44; p = 0.002) were features that outweighed the others when diagnosing EC. The diagnostic model based on these features showed good discrimination ability in both the derivation set (AUC = 0.888, 95% confidence interval [95% CI] = 0.825–0.950) and validation set (AUC = 0.902, 95% CI = 0.832–0.972). CONCLUSION: More-than-half-depth of involvement of the dermal layer, “submarine sign,” and morphology are relatively better US features than the others for diagnosing EC.