Wang, Ren, Wang, Li, Zhu, Shan, Wang, Zhao, Li, Ren, Guo, Xu, and Sun: Role of high-frequency ultrasound in differentiating benign and malignant skin lesions: potential and limitations



This study examined the diagnostic value of high-frequency ultrasound (HFUS) features in differentiating between benign and malignant skin lesions.


A total of 1,392 patients with 1,422 skin lesions who underwent HFUS examinations were included in an initial dataset (cohort 1) to identify features indicative of malignancy. Qualitative clinical and HFUS characteristics were recorded for all lesions. To determine which HFUS and clinical features were suggestive of malignancy, univariable and multivariable logistic regression analyses were employed. The diagnostic performance of HFUS features combined with clinical information was evaluated. This assessment was validated using internal data (cohort 2) and multicenter external data (cohort 3).


Features significantly associated with malignancy included age above 60 years; lesion location in the head, face, and neck or genital regions; changes in macroscopic appearance; crawling or irregular growth pattern; convex or irregular base; punctate hyperechogenicity; blood flow signals; and feeding arteries. The area under the receiver operating characteristic curve, sensitivity, and specificity of HFUS features combined with clinical information were 0.946, 92.5%, and 86.9% in cohort 1; 0.870, 93.1%, and 80.8% in cohort 2 (610 lesions); and 0.864, 86.2%, and 86.6% in cohort 3 (170 lesions), respectively. However, HFUS is not suitable for evaluating lesions less than 0.1 mm in thickness or lesions exhibiting surface hyperkeratosis.


In a clinical setting, the integration of HFUS with clinical information exhibited good diagnostic performance in differentiating malignant and benign skin lesions. However, its utility was limited in evaluating extremely thin lesions and those exhibiting hyperkeratosis.

Graphic Abstract


The skin serves as a physical barrier against the external environment, resulting in a high incidence of malignant and benign skin lesions. These conditions impose a substantial socioeconomic burden and reduce patient quality of life [1]. Differentiating between malignant and benign skin lesions is crucial for determining an appropriate treatment protocol and improving health outcomes [2,3]. While biopsy is considered the gold standard for diagnosing skin lesions, it is an invasive procedure with cosmetic concerns. Furthermore, biopsy specimens do not provide information about the spatial relationships between the lesion and adjacent structures [4]. Consequently, ongoing efforts are aimed at diagnosing skin lesions using noninvasive methods. These approaches have the potential to aid clinicians in the early detection, diagnosis, and management of skin lesions [5,6].
Noninvasive diagnostic techniques such as dermatoscopy and optical coherence tomography have inherent depth constraints, rendering them unsuitable for lesions located below the papillary dermis [6,7]. Computed tomography and magnetic resonance imaging exhibit limited spatial resolution and are useful only in identifying skin lesions larger than 5 mm [8,9]. Therefore, those noninvasive techniques are not sufficient for the evaluation of skin lesions. High-frequency ultrasound (HFUS; 15-50 MHz) provides a superior balance of penetration depth and spatial resolution, enabling the clear identification of skin layers [4,10]. HFUS is particularly valuable due to its capacity to provide additional anatomical details, such as depth, contour, and vascularization [11,12].
Ultrasonography (US) was initially proposed for skin measurement in the late 1970s. However, due to frequency limitations, this imaging tool was not used to diagnose skin cancer until the early 1990s [5]. Recently, with increases in transducer frequency, US has become more widely applied in dermatology [4,5,13-19]. Consequently, guidelines for performing dermatologic US have been established [10,12]. Nevertheless, research has largely been limited to describing and comparing US features of specific skin diseases [14,19-24], such as differentiating invasive basal cell carcinoma (BCC) from squamous cell carcinoma and Bowen disease from superficial BCC. However, given the vast array of skin diseases, an understanding of US features based on a limited number of conditions is insufficient for generalization. Therefore, this study aimed to identify and evaluate universal US features across a broader spectrum of skin diseases to improve the identification of benign and malignant skin lesions.

Materials and Methods

Compliance with Ethical Standards

The study was registered at (No. ChiCTR 2300068635). Institutional review board approval was obtained from Shanghai Tenth People’s Hospital (No. SHSY-IEC-5.0/22K297/P01) and Shanghai Skin Disease Hospital (No. SSDH-IEC-SG-029-4.1). Informed consent was waived for the patients in cohort 1 and cohort 2. For patients in cohort 3, informed consent was obtained.


The inclusion criteria were as follows: (1) pathologic results for the lesion were available after surgical resection; and (2) complete clinical data and HFUS images were accessible for the patient. The exclusion criteria encompassed cases in which: (1) the pathologic result was indefinite; (2) the lesion had been treated (including surgery, laser, photodynamic therapy, or radiotherapy) prior to HFUS examination; or (3) the quality of the HFUS image was poor (Fig. 1).
From March 2017 to December 2021, this study retrospectively included consecutive patients who underwent HFUS examinations for skin lesions at Shanghai Skin Disease Hospital and Shanghai Tenth People’s Hospital. These patients were randomly divided into two groups (cohort 1 and cohort 2) using a 7:3 ratio. Informed consent was waived for the patients in these cohorts. Subsequently, from January 2022 to June 2022, data were prospectively collected from patients at multiple centers (Jiading District Central Hospital affiliated Shanghai University of Medicine & Health Sciences, Shanghai, China; Maanshan Hospital, Maanshan, China; Jinshan Hospital, Fudan University, Shanghai, China; and Zhongshan Hospital, Fudan University, Shanghai, China) to serve as an external validation dataset (cohort 3). For patients in cohort 3, informed consent was obtained.
The basic clinical data collected included age, sex, lesion location—categorized as head, face, and neck; trunk; genital region; or limbs, and changes in macroscopic appearance. The last of these included variations in pigmentation, ulceration, bleeding, and keratinization, as documented through photographs captured with a mobile phone or derived from electronic medical records.

HFUS Examinations

Each lesion was assessed using a high-frequency linear-array transducer, with the specific device varying across centers. The frequencies of these transducers were 9-33 MHz (Aplio i800, Canon Medical Systems, Tochigi, Japan), 10-22 MHz (MyLab TMC class C, Esaote, Genoa, Italy), 6-20 MHz (Aixplorer, Supersonic Imagine, Aix-en-Provence, France), 22-38 MHz (Paragon, Kolo Medical, Suzhou, China), and 50 MHz (MD-300SII, Meda Co., Ltd., Tianjin, China). The operator selected the appropriate frequency based on the depth of the lesion.
In cohorts 1 and 2, HFUS examinations were conducted by four radiologists (L.G., D.S., W.R., and A.Z.), each with 5-8 years of experience in dermatologic US. In cohort 3, HFUS examinations were carried out by a different set of four radiologists (L.W., D.L., T.T.R., and J.W.), who also had 5-8 years of experience in dermatologic US.

Imaging Analysis

The images were analyzed using B-mode HFUS to identify various features, as outlined in prior studies [14,19,20-23]. These features included layer involvement (affecting one layer or multiple layers); growth pattern (categorized as nodular if the lesion exhibited regular morphology, crawling if it grew parallel to the skin, or irregular if it displayed a lobulated or other irregular morphology); surface characteristics (flat if the lesion’s surface is level with the surrounding skin, convex with a circumscribed boundary if it is raised above the surrounding skin, concave equivalent to ulceration if it is depressed below the skin, or irregular if the skin surface of the lesion displays sags and crests); base shape (flat if the lesion’s base is straight, convex if it is semicircular and convex to deeper tissue, or irregular if it is lobulated, jagged, or sharply angled); margin definition (well-defined or ill-defined); the status of the stratum corneum (normal if it appears as a continuous, smooth, and fine linear hyperechoic band; abnormal keratosis including hyperkeratosis if the hyperechoic band is thicker, rough, or wrinkled; and parakeratosis or detachment of the stratum corneum if the hyperechoic band is interrupted); echogenicity (hyperechoic, isoechoic, hypoechoic, or anechoic); internal homogeneity (homogeneous or heterogeneous); internal hyperechogenicity (absent, punctate, or strip/patchy); cystic components (absent or present); and posterior acoustic features (normal, attenuation, or enhancement). Additionally, the presence of blood flow signals and feeding arteries was documented using color Doppler US. Definitions of the HFUS features and corresponding schematic diagrams are provided in the Supplementary Text 1 and Supplementary Fig. 1.
All HFUS images were evaluated by two independent radiologists from different centers (Q.W. and X.L., each with 8 years of experience in dermatologic US), who were blinded to the pathologic results and clinical data. Before reviewing the HFUS images, the radiologists convened to establish a consensus on the HFUS features mentioned earlier in this paper by examining 200 randomly selected cases from cohort 1. Any discordance between the radiologists was resolved through discussion until a consensus was reached. The results were then utilized for further analysis.

Reproducibility of HFUS Features

An additional 200 cases were randomly selected from cohort 1 for interobserver reproducibility analysis. Both radiologists independently reviewed all images.

Statistical Analysis

Quantitative data were described as the mean±standard deviation when they followed a normal distribution. The chi-square test or Fisher exact test was used to compare categorical variables, while an independent two-sample t-test was employed for continuous variables. In cohort 1, univariate analysis was conducted to assess the correlation between predictive factors and malignant skin lesions. Multivariate logistic regression analysis was then utilized to identify significant independent predictors of malignancy. During this analysis, all nonsignificant variables were sequentially removed. The area under the receiver operating characteristic curve (AUROC), along with sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy, were evaluated. The Cohen kappa test was applied to assess observer reproducibility in the evaluation of HFUS features. Data analysis was performed using SPSS version 20.0 (IBM Corp., Armonk, NY, USA).


Patient Characteristics

From March 2017 to December 2021, a total of 3,140 localized skin lesions in 3,067 consecutive patients were examined using HFUS, with 972 lesions subsequently excluded. The remaining 2,168 skin lesions in 2,110 patients were selected for further analysis. Upon review, 64 lesions were not visible on HFUS, and 72 lesions could not be assessed by HFUS due to posterior acoustic attenuation (Fig. 1).
Among the 2,032 assessable skin lesions from 1,990 patients, 1,422 lesions from 1,392 patients (631 women and 761 men; mean age, 58.5±20.3 years; range, 3 to 98 years) were assigned to cohort 1. Within this cohort, the final diagnoses covered 74 entities, including benign skin tumors, inflammatory and infectious diseases, and malignant skin tumors. Both benign skin tumors and inflammatory and infectious diseases were categorized as benign skin lesions (Fig. 2). The remaining 610 skin lesions from 598 patients were allocated to cohort 2.
From January 2022 to June 2022, 170 skin lesions in 155 patients were included in cohort 3. The baseline characteristics of the three cohorts are presented in Table 1.

HFUS and Clinical Features

In univariate analysis, the following clinical data were significantly associated with malignancy: age over 60 years, lesion location, and changes in macroscopic appearance (all P<0.001). On HFUS, malignancy was significantly associated with features including layer involvement, growth pattern, surface characteristics, base shape, margin definition, the status of the stratum corneum, echogenicity, internal hyperechogenicity, posterior acoustic features, blood flow signals, and feeding arteries (Supplementary Table 1). Multivariable logistic regression analysis results indicated that significant predictors of malignant skin lesions included age over 60 years, lesion location (in the head, face, and neck or genital regions), changes in macroscopic appearance, crawling or irregular growth pattern, convex or irregular base, punctate hyperechogenicity, blood flow signals, and feeding arteries were significant predictors of malignant skin lesions (Table 2, Figs. 3, 4).

Diagnostic Performance of HFUS Features

In cohort 1, the AUROC for HFUS combined with clinical data was 0.946 (95% confidence interval [CI], 0.935 to 0.957). The sensitivity, specificity, PPV, NPV, and accuracy were 92.5%, 86.9%, 84.8%, 93.6%, and 89.4%, respectively.
The diagnostic performance of HFUS features in identifying malignant skin lesions was evaluated with both internal and external datasets. The AUROC, along with sensitivity, specificity, PPV, NPV, and accuracy, were assessed. In cohort 2, the AUROC was 0.870 (95% CI, 0.839 to 0.900), with a sensitivity of 93.1%, specificity of 80.8%, PPV of 80.1%, NPV of 93.4%, and accuracy of 86.4%. In cohort 3, the AUROC was 0.864 (95% CI, 0.801 to 0.927), with a sensitivity of 86.2%, specificity of 86.6%, PPV of 76.9%, NPV of 92.4%, and accuracy of 86.5%.
The present findings indicate that HFUS, when combined with clinical data, demonstrated good diagnostic performance. Notably, however, certain cases warrant further discussion. For instance, some lesions exhibited hyperkeratosis in the stratum corneum, which limited assessment with HFUS due to the posterior acoustic attenuation caused by the hyperkeratosis (Fig. 5). Furthermore, the resolution of HFUS is insufficient to detect some extremely thin lesions, such as port wine stains, dysplastic nevi, and pigment deposits (including melanin, tattoo inks, or hemosiderin) that exhibit macroscopic changes (Fig. 6).
Malignant melanoma (MM) is the most aggressive form of skin tumor and generally has a poor prognosis. The present findings indicate that MM can exhibit a range of morphologies when observed using HFUS. These include the absence of findings (n=17), irregular masses (n=8), and nodular masses (n=12).

Interobserver Reproducibility of HFUS Features

Interobserver reproducibility analysis revealed a substantial to near-perfect agreement on significant HFUS features (Supplementary Table 2).


HFUS is an important radiological tool for assessing skin diseases. Combining HFUS with clinical information yielded valuable insights for differentiating between benign and malignant skin lesions.
To maximize the information provided by HFUS, operators of dermatologic US must understand which HFUS features are significant for diagnosing malignant skin tumors. The DERMUS team, an international working group, has proposed the establishment of a database to facilitate the generation of standardized US findings for a broad spectrum of dermatologic conditions [10].
Routine US findings include echogenicity, margin, morphology, internal homogeneity, components, and size. A previous study of 4,338 skin lesions demonstrated that US improved the accuracy of clinical diagnosis from 73% to 97%, with an overall sensitivity of 99% and a specificity of 100% [4]. These results underscore the value of US in the diagnosis of skin lesions. However, this large retrospective study only listed certain routine US features, without providing a thorough evaluation. Furthermore, the study did not detail other US features, which may be attributed to the limited frequency range used in the research (only 7-15 MHz).
Beyond routine US features, previous studies have demonstrated that characteristics such as the basal border, surface features, and stratum corneum features observed on HFUS are valuable for differentiating between skin diseases [14,19,21-23]. These HFUS characteristics are specific to skin diseases and play a key role in their diagnosis. Consequently, in the present study, several high-frequency transducers were employed to examine skin lesions at various depths to more thoroughly observe the HFUS findings.
In this study, a relatively large number of cases were analyzed to identify significant HFUS and clinical features associated with skin lesion malignancy. The findings indicate that a crawling or irregular growth pattern, a convex or irregular base, punctate hyperechogenicity, the presence of blood flow signals, and the observation of feeding arteries are significant HFUS predictors of malignancy in skin lesions.
Regarding the growth patterns of skin lesions, an irregular pattern is associated with aggressive malignant tumors in other organs [25-27]. The present results indicated that this feature is also applicable to malignant skin tumors. However, a crawling growth pattern has been mentioned exclusively for skin lesions and has frequently been observed in conditions such as actinic keratosis, Bowen disease, extramammary Paget disease, and superficial basal cell carcinomas [14,19,22]. Regarding the base of the lesion, a regular basal border is generally considered a noninvasive characteristic [28]. In contrast, skin lesions such as squamous cell carcinoma, with convex or irregular basal borders, have been confirmed to extend into the underlying tissues on histopathology [14]. Additionally, the present study identified punctate hyperechogenicity as a predictor of malignant skin lesions. Previous studies have demonstrated that punctate hyperechogenicity is a characteristic feature of BCC, corresponding to calcium deposits, keratinized cells, or prominent basal cell aggregates [12,19,29,30].
The use of feeding arteries as a predictor of malignant skin tumors has not been directly supported by evidence from previous studies. Nevertheless, the present findings, which are based on a relatively large case series, support their predictive value. Similarly, another study indicated that abnormal angioarchitecture, characterized by features such as hypervascularization, disorganized blood flow, and the presence of multiple vascular pedicles, may be indicative of a malignant lesion [11].
Unlike lesions in other organs, in addition to internal information evident on imaging, skin lesions provide specific clinical data such as location and macroscopic appearance [24,31]. These features cannot be blinded for operators during HFUS examinations. For instance, the HFUS features of MM can vary widely, from being undetectable to presenting as a nodular mass. However, melanin pigmentation is often observed. Consequently, relying solely on HFUS may result in missed MM lesions. Therefore, for the diagnosis of skin lesions, HFUS should be used in conjunction with clinical information to better reflect real-world clinical scenarios. The present findings indicate that age above 60 years, lesion location in the head, face, and neck or genital regions, and changes in macroscopic appearance are significant clinical predictors of malignant skin lesions.
Most malignant skin tumors arise in the head, face, and neck region, which is most frequently exposed to the sun. This is attributed to ultraviolet light being the primary carcinogen, capable of inducing genetic mutations [32]. A prior study demonstrated that variations in macroscopic appearance were associated with cancers of different levels of aggressiveness, with malignant forms resulting in more tissue loss compared to benign lesions [14]. Consequently, in line with the present findings, these clinical features could represent valuable indicators for HFUS operators.
In the present study, the combination of HFUS features and clinical data demonstrated good diagnostic performance, aligning with findings from a prior study [4]. Furthermore, the results were confirmed through validation with both internal and external datasets, underscoring the practicality of these features. However, the limitations of HFUS should be acknowledged. Specifically, the characterization of certain lesions with hyperkeratosis using HFUS was challenging. Moreover, lesions that were extremely thin—less than 0.1 mm thick—were not detectable on HFUS. Consequently, additional examinations or biopsy procedures may be required for such lesions.
This study had several limitations. First, despite the inclusion of many lesions and conditions (1,422 across 74 entities), a greater diversity of skin diseases is encountered in clinical practice. The limited sample size and the range of skin entities included may have introduced selection bias, particularly for rare skin conditions. Second, HFUS is operator-dependent, and the interobserver reproducibility analysis was performed on only a subset of the cases. Finally, the findings are based on a Chinese population, which predominantly represents a specific and limited range of skin types.
Consequently, prospective international studies including all skin types are required.
In summary, HFUS is an important noninvasive diagnostic tool for the diagnosis of skin lesions in clinical practice. However, its utility is limited when evaluating extremely thin lesions or those with hyperkeratosis. In actual clinical scenarios, integrating HFUS with clinical information demonstrates good diagnostic performance.


Author Contributions

Conceptualization: Wang Q, Ren W, Zhao Y, Guo L, Xu H, Sun L. Data acquisition: Wang Q, Zhu A, Shan D, Wang J, Li D, Ren TT. Data analysis or interpretation: Wang Q, Ren W, Wang L, Li X. Drafting of the manuscript: Wang Q. Critical revision of the manuscript: Ren W, Wang L, Li X, Zhu A, Shan D, Wang J, Zhao Y, Li D, Ren TT, Guo L, Xu H, Sun L. Approval of the final version of the manuscript: all authors.

Conflict of Interest

No potential conflict of interest relevant to this article was reported.

Supplementary Material

Supplementary Table 1.

Association between malignant skin lesions and various HFUS features (

Supplementary Table 2.

Interobserver reproducibility of HFUS features (

Supplementary Fig. 1.

HFUS features and the corresponding schematic diagrams for skin lesions (

Supplementary Text 1.

HFUS features and their definitions for skin lesions (


1. GBD 2017 DALYs and HALE Collaborators. Global, regional, and national disability-adjusted life-years (DALYs) for 359 diseases and injuries and healthy life expectancy (HALE) for 195 countries and territories, 1990-2017: a systematic analysis for the Global Burden of Disease Study 2017. Lancet 2018;392:1859–1922.
pmid pmc
2. Nasr I, McGrath EJ, Harwood CA, Botting J, Buckley P, Budny PG, et al. British Association of Dermatologists guidelines for the management of adults with basal cell carcinoma 2021. Br J Dermatol 2021;185:899–920.
crossref pmid pdf
3. Alam M, Armstrong A, Baum C, Bordeaux JS, Brown M, Busam KJ, et al. Guidelines of care for the management of cutaneous squamous cell carcinoma. J Am Acad Dermatol 2018;78:560–578.
pmid pmc
4. Wortsman X, Wortsman J. Clinical usefulness of variable-frequency ultrasound in localized lesions of the skin. J Am Acad Dermatol 2010;62:247–256.
crossref pmid
5. Kleinerman R, Whang TB, Bard RL, Marmur ES. Ultrasound in dermatology: principles and applications. J Am Acad Dermatol 2012;67:478–487.
crossref pmid
6. Schneider SL, Kohli I, Hamzavi IH, Council ML, Rossi AM, Ozog DM. Emerging imaging technologies in dermatology: part I: basic principles. J Am Acad Dermatol 2019;80:1114–1120.
7. Oh BH, Kim KH, Chung KY. Skin imaging using ultrasound imaging, optical coherence tomography, confocal microscopy, and two-photon microscopy in cutaneous oncology. Front Med (Lausanne) 2019;6:274.
crossref pmid pmc
8. Antoch G, Vogt FM, Freudenberg LS, Nazaradeh F, Goehde SC, Barkhausen J, et al. Whole-body dual-modality PET/CT and whole-body MRI for tumor staging in oncology. JAMA 2003;290:3199–3206.
crossref pmid
9. Expert Panel on Thoracic Imaging, Ackman JB, Chung JH, Walker CM, Bang TJ, Carter BW, et al. ACR Appropriateness Criteria(R) Imaging of Mediastinal Masses. J Am Coll Radiol 2021;18(5S):S37–S51.
10. Wortsman X, Alfageme F, Roustan G, Arias-Santiago S, Martorell A, Catalano O, et al. Guidelines for performing dermatologic ultrasound examinations by the DERMUS group. J Ultrasound Med 2016;35:577–580.
crossref pmid pdf
11. Crisan D, Badea AF, Crisan M, Rastian I, Gheuca Solovastru L, Badea R. Integrative analysis of cutaneous skin tumours using ultrasonogaphic criteria. Preliminary results. Med Ultrason 2014;16:285–290.
12. Alfageme F, Wortsman X, Catalano O, Roustan G, Crisan M, Crisan D, et al. European Federation of Societies for Ultrasound in Medicine and Biology (EFSUMB) position statement on dermatologic ultrasound. Ultraschall Med 2021;42:39–47.
crossref pmid
13. Li MX, Wang Q, Li XL, Zhao CK, Zhu RZ, Chen J, et al. Imaging findings of Bowen's disease: a comparison between ultrasound biomicroscopy and conventional high-frequency ultrasound. Skin Res Technol 2020;26:654–663.
crossref pmid pdf
14. Zhu AQ, Wang LF, Li XL, Wang Q, Li MX, Ma YY, et al. High-frequency ultrasound in the diagnosis of the spectrum of cutaneous squamous cell carcinoma: noninvasively distinguishing actinic keratosis, Bowen's disease, and invasive squamous cell carcinoma. Skin Res Technol 2021;27:831–840.
crossref pmid pdf
15. Zattar LC, Alves ID, Bernal E, Zattar GA, Marcelino AS, Cerri LM, et al. Radiologic approach for skin US. Radiographics 2022;42:E73–E74.
crossref pmid
16. Catalano O, Wortsman X. Dermatology ultrasound: imaging technique, tips and tricks, high-resolution anatomy. Ultrasound Q 2020;36:321–327.
crossref pmid
17. Wortsman X. Top advances in dermatologic ultrasound. J Ultrasound Med 2023;42:521–545.
crossref pmid pdf
18. Wortsman X. Top applications of dermatologic ultrasonography that can modify management. Ultrasonography 2023;42:183–202.
crossref pmid pmc pdf
19. Wang LF, Zhu AQ, Wang Q, Li XL, Yan JN, Hui S, et al. Value of high-frequency ultrasound for differentiating invasive basal cell carcinoma from non-invasive types. Ultrasound Med Biol 2021;47:2910–2920.
crossref pmid
20. Xu HX, Guo LH, Zhu AQ, Shan DD, Wang Q. Terminology, image Interpretation, and artifacts for skin ultrasound. In: Xu H, Guo L, Wang Q, eds. Diagnostic ultrasound in dermatology. Singapore: Springer Singapore, 2022:55–69.

21. Zhang YQ, Wang LF, Ni N, Li XL, Zhu AQ, Guo LH, et al. The value of ultra-high-frequency ultrasound for the differentiation between superficial basal cell carcinoma and Bowen's disease. Dermatology 2023;239:572–583.
crossref pmid pdf
22. Shan D, Wu N, Wang Q, Ren W, Zhu A, Wang L, et al. Value of pseudopod sign on high-frequency ultrasound in predicting the pathological invasion of extramammary Paget's disease lesions. J Eur Acad Dermatol Venereol 2022;36:1235–1245.
crossref pmid pdf
23. Qin J, Wang J, Zhu Q, Liu J, Gao Y, Wang Y, et al. Usefulness of high-frequency ultrasound in differentiating basal cell carcinoma from common benign pigmented skin tumors. Skin Res Technol 2021;27:766–773.
crossref pmid pdf
24. Chen ZT, Yan JN, Zhu AQ, Wang LF, Wang Q, Li L, et al. High-frequency ultrasound for differentiation between high-risk basal cell carcinoma and cutaneous squamous cell carcinoma. Skin Res Technol 2022;28:410–418.
crossref pmid pmc pdf
25. Ruan J, Xu X, Cai Y, Zeng H, Luo M, Zhang W, et al. A practical CEUS thyroid reporting system for thyroid nodules. Radiology 2022;305:149–159.
crossref pmid
26. American College of Radiology. Breast Imaging Reporting and Data System, breast imaging atlas. 4th ed. Reston, VA: American College of Radiology, 2003.

27. Kwak JY, Han KH, Yoon JH, Moon HJ, Son EJ, Park SH, et al. Thyroid imaging reporting and data system for US features of nodules: a step in establishing better stratification of cancer risk. Radiology 2011;260:892–899.
crossref pmid
28. Crisan M, Crisan D, Sannino G, Lupsor M, Badea R, Amzica F. Ultrasonographic staging of cutaneous malignant tumors: an ultrasonographic depth index. Arch Dermatol Res 2013;305:305–313.
crossref pmid pdf
29. Uhara H, Hayashi K, Koga H, Saida T. Multiple hypersonographic spots in basal cell carcinoma. Dermatol Surg 2007;33:1215–1219.
crossref pmid
30. Wortsman X, Vergara P, Castro A, Saavedra D, Bobadilla F, Sazunic I, et al. Ultrasound as predictor of histologic subtypes linked to recurrence in basal cell carcinoma of the skin. J Eur Acad Dermatol Venereol 2015;29:702–707.
crossref pmid pdf
31. Bayan CY, Khanna T, Rotemberg V, Samie FH, Zeitouni NC. A review of non-invasive imaging in extramammary Paget's disease. J Eur Acad Dermatol Venereol 2018;32:1862–1873.
crossref pmid pdf
32. Peris K, Fargnoli MC, Garbe C, Kaufmann R, Bastholt L, Seguin NB, et al. Diagnosis and treatment of basal cell carcinoma: European consensus-based interdisciplinary guidelines. Eur J Cancer 2019;118:10–34.
crossref pmid

Flowchart of patient selection and the number of skin lesions.

Lesions exhibiting hyperkeratosis could not be evaluated with highfrequency ultrasound (HFUS). This category includes diseases such as warts, seborrheic keratoses, squamous cell carcinomas, Bowen disease, and actinic keratoses, among others. Examples of lesions that are not detectable on HFUS include certain nevi of the nail matrix, dysplastic nevi, and port wine stains.
Fig. 1.

A schematic illustration of the skin entities included in the training set.

The set comprised 1,422 lesions, represented by the gray area, and encompassed three categories: benign skin tumors (green area), inflammatory and infectious diseases (yellow area), and malignant skin tumors (red area). The size of each circle is proportional to the sample size within that category. Both benign skin tumors and inflammatory and infectious diseases were classified as benign skin lesions.
Fig. 2.

A case of benign skin lesion.

A. Visual observation reveals a black papule (arrows) near the left ear of a 56-year-old woman. B. High-frequency ultrasound (HFUS; frequency of 15 MHz) reveals an ill-defined lesion (arrows) in the epidermis and dermis. C. Ultra-HFUS (frequency of 34 MHz) indicates that the lesion is located in the dermis with posterior acoustic attenuation (arrows). D. Color Doppler ultrasound shows no blood flow signals within the lesion (arrows). E. Histological analysis confirms the diagnosis of a nevus (H&E, ×20).
Fig. 3.

A case of malignant skin lesion.

A. Visual observation reveals a reddish bump (arrows) on the left wrist of an 83-year-old woman, characterized by a concave center and a prominent rim. B. High-frequency ultrasound (HFUS; frequency of 15 MHz) shows an irregular lesion (arrows) in the epidermis and dermis, featuring an irregular base and hyperkeratosis. C. Color Doppler ultrasound demonstrates blood flow signals within the lesion (arrows) and a feeding artery (arrowheads). D. Ultra-HFUS (frequency of 34 MHz) reveals punctate hyperechogenicity (arrowheads) within the lesion (arrows). E. Histological analysis confirms the diagnosis of squamous cell carcinoma (H&E, ×20).
Fig. 4.

A case of skin lesion that can't be evaluated by high frequency ultrasound.

A. Visual observation reveals a dark brown papule (arrows) on the face of a 78-year-old woman. B. The lesion could not be evaluated with ultra-high frequency ultrasound (frequency of 34 MHz) due to posterior acoustic attenuation resulting from hyperkeratosis (arrows). The lesion was confirmed to be seborrheic keratosis.
Fig. 5.

A case of skin lesion that is invisible on high frequency ultrasound.

A. Visual observation reveals erythema (arrows) behind the ear of a 34-year-old woman. B. The lesion is not visible on ultra-high frequency ultrasound (frequency of 34 MHz). C. Color Doppler ultrasound also shows no abnormalities. The lesion was confirmed to be a port wine stain. e, epidermis; d, dermis; st, subcutaneous tissue.
Fig. 6.
Table 1.
Baseline data of cohorts 1, 2, and 3
Cohort 1 (n=1,422)
P-value Cohort 2 (n=610)
P-value Cohort 3 (n=170)
Benign lesions (n=793) Malignant lesions (n=629) Benign lesions (n=334) Malignant lesions (n=276) Benign lesions (n=112) Malignant lesions (n=58)
No. of patients 768 624 322 276 99 56
Sex 0.05 0.088 0.457
 Male 438 323 195 147 48 25
 Female 330 301 127 129 51 31
Age (year) 47.3±18.2 72.8±11.9 <0.001 46.4±18.6 72.9±11.0 <0.001 50.0±19.1 70.1±14.2 <0.001
Location <0.001 <0.001 0.006
 Head, face, and neck 321 389 135 178 45 33
 Genital region 32 101 12 35 2 4
 Limbs 197 59 80 26 40 8
 Trunk 243 80 107 37 25 13
Changes in macroscopic appearance <0.001 <0.001 <0.001
 Present 344 622 109 274 50 55
 Absent 449 7 225 2 62 3
Thickness (mm) 7.0±5.6 4.8±4.2 <0.001 7.1±6.1 4.0±3.7 <0.001 6.5±4.8 4.2±3.9 0.002

Values are presented as number of lesions or mean±standard deviation.

Table 2.
Results of multivariate analysis in predicting malignant skin lesions
Parameter B SE OR (95% CI) P-value
Age≥60 years 2.467 0.217 11.784 (7.704-18.026) <0.001
 Head, face, and neck 1.378 0.289 3.969 (2.250-6.999) <0.001
 Genital region 1.699 0.409 5.305 (2.381-11.822) <0.001
 Trunk 0.596 0.329 1.814 (0.953-3.454) 0.070
 Limb NA NA 1 N/A
Changes in macroscopic appearance 2.479 0.470 11.928 (4.748-29.968) <0.001
Growth pattern
 Nodular NA NA 1 N/A
 Crawling 2.893 0.323 18.047 (9.579-34.001) <0.001
 Irregular 1.374 0.311 3.953 (2.148-7.274) <0.001
Base shape
 Flat NA NA 1 N/A
 Convex 0.964 0.281 2.623 (1.513-4.548) 0.001
 Irregular 2.253 0.406 9.518 (4.294-21.101) <0.001
 Absent NA NA 1 N/A
 Punctate 1.313 0.236 3.716 (2.342-5.897) <0.001
 Strip/patchy -1.737 0.754 0.176 (0.040-0.771) 0.021
Color Doppler flow signal 1.805 0.268 6.079 (3.596-10.276) <0.001
Feeding arteries 1.245 0.407 3.472 (1.563-7.714) 0.002

SE, standard error; OR, odds ratio; CI, confidence interval; NA, not applicable.

Editorial Office
A-304 Mapo Trapalace, 53 Mapo-daero, Mapo-gu, Seoul 04158, Korea
TEL : +82-2-763-5627   FAX : +82-2-763-6909   E-mail :
About |  Browse Articles |  Current Issue |  For Authors and Reviewers
Copyright © Korean Society of Ultrasound in Medicine.                 Developed in M2PI
Zoom in Close layer