The value of color-map virtual touch tissue imaging (CMV) in the differential diagnosis of benign and malignant breast lesions

Abstract

BACKGROUND:

Researchers have evaluated the virtual touch tissue imaging (VTI) value in the diagnosis of breast lesions, mostly based on gray-scale.

PURPOSE:

This study aimed to evaluate the value of color-map virtual touch tissue imaging (CMV) in the diagnosis of breast lesions.

METHODS:

We retrospectively analyzed the VTI images of 55 breast lesions in 49 female patients who underwent an examination of breast lesions in our hospital from January 2019 to December 2019. The pathological results were taken as the gold standard. The receiver operating characteristic (ROC) curve of CMV was analyzed, and its diagnostic performance was evaluated. Weighted Kappa (k) statistics were used to assess the inter-observer agreement for CMV.

RESULTS:

A total of 55 breast lesions were included, including 19 malignant lesions and 36 benign lesions. Multivariate analysis showed that patients with higher CMV scores (P = 0.014, odds ratio [OR] = 13.667, 95% confidence interval = 1.702–109.773) were independent predictors of breast cancer. The sensitivity, specificity, and the area under curve (AUC) of CMV were 94.47%, 72.22%, and 0.912. The CMV’s inter-observer agreement was almost perfect among radiologists with different work experience (k = 0.854, standard error = 0.049, 95% CI = 0.758–0.950).

CONCLUSIOS:

CMV has high accuracy and repeatability in the diagnosis of malignant breast lesions.

Keywords

Breast lesion ultrasound elastography

1 Introduction

So far, mammography is the only technique recommended by the guidelines for screening malignant breast lesions. However, in patients with high-density breast tissue, its diagnostic accuracy is limited [1]. The appearance of ultrasonic elastography provides an effective method to improve the diagnosis of malignant breast lesions.

Ultrasonic elastography is a kind of ultrasonic technology that measures tissue deformation under pressure and shows tissue hardness [2, 3]. In the breast, radiologists have applied elastography to breast nodules, which depends on the fact that most cancers are more rigid than normal breast and benign lesions [4 –9].

Compared with strain elasticity, acoustic radiation force impulse (ARFI) is more objective. The ultrasonic instrument uses a short-lasting acoustic radiation force instead of external compression, resulting in local displacement in the tissue. Virtual touch tissue imaging (VTI) is a qualitative response of ARFI. VTI images depict relative rigidity with gray-scale or color-maps [10 –12].

Researchers have evaluated VTI’s value in breast lesions’ characterization in several studies, mostly based on gray-scale VTI (GSV) [13 –17]. However, the human eyes’ ability to distinguish different gray-scale is much weaker than that of different colors.

Therefore, the purpose of this study was to evaluate the value of color-map virtual touch tissue imaging (CMV) in the diagnosis of breast lesions.

2 Materials and methods

Our institutional ethics committee has approved this retrospective study, and informed consent was abandoned.

2.1 Patients

From January 2019 to December 2019, we included all consecutive patients who underwent conventional ultrasound (US) and ARFI examinations of breast nodules in our hospital. Patients were excluded from the study if US data or pathological diagnosis were incomplete.

2.2 Instruments

This study used the ACUSON S2000 US scanner system (Siemens Medical Solutions, Mountain View, CA, USA) and the 9L4 linear probe for standard conventional US and ARFI examination.

2.3 Image acquisition

A radiologist performed both breast ultrasound and VTI examinations with ten years of ultrasound experience to avoid differences between operators. First of all, a conventional US examination was conducted on each patient. In the application of VTI, the region of interest(ROI) was adjusted to include the entire lesion (70% of the total ROI) and some surrounding normal breast tissue (20% of the real ROI). For large nodules whose volume exceeded the ROI limit, ROI was placed at the junction between the nodule and the surrounding breast tissue. Finally, ROI included both partial lesions and some surrounding breast tissue. The elastic distribution of tissue was calculated, and the computed results showed a gray-scale image or color-coding image. In the GSV, the dark area indicates that the tissue hardness is high, and the bright spot suggests that the tissue hardness is low. In CMV, red and orange indicate high hardness, green indicates medium hardness, and blue and purple indicate low hardness. In our study, all cases were examined by CMV and GSV at the same time. All conventional US and VTI images were stored and further analyzed.

2.4 Image analysis

Two independent radiologists (5 years of experience) retrospectively analyzed VTI images. Two radiologists did not participate in VTI image acquisition. They did not know the patients’ clinical information, pathological results, and other imaging results. After the identification information was hidden, individual cases were presented in random order. When there was a discrepancy in the two radiologists’ assessments, the third radiologist (more than 15 years of experience) reviewed the image and decided.

The CMV score was evaluated according to hue. The lesions were divided into four categories according to the criteria of Zhou et al. [18]: Score 1, the primary color of the lesion is green, blue, or purple (low hardness); score 2, the most extensive range of color of the lesion is orange, the secondary color is green, and a small number of areas of other colors (moderate hardness); score 3, red or orange occupies most of the lesion, with only a few different colors (high hardness); score 4, the color of the lesion and its surrounding tissue is mainly red or orange, and only a few other colors (high hardness of the lesion and its surrounding tissue) (Fig. 1).

Fig. 1

Color-map virtual touch tissue imaging scores of the breast lesions: a score 1; b score 2; c score 3; d score 4.

Four weeks after the completion of the CMV image analysis, GSV image analysis was performed. Two independent radiologists scored the GSV of breast lesions according to Xu’s 6-point scoring system [13]: Score 1, prominent white; score 2, mainly white, a small amount of black; score 3, black and white equal portions; score 4, primarily black, with a few white; score 5, almost entirely black; score 6, completely black (Fig. 2).

Fig. 2

Gray-scale virtual touch tissue imaging scores of the breast lesions: a score 1; b score 2; c score 3; d score 4; e score 5; f score 6.

2.5 Inter-observer agreement for VTI elastography

The inter-observer agreement was carried out by two other radiologists who evaluated all VTI images of the study on the same day, based on the above VTI elasticity score. Two radiologists had different ultrasound experience (2 years and 5 years).

2.6 Statistical analyses

This study used SPSS version 20.0 and MedCalc version 18.0 software for statistical analysis. The quantitative data were expressed by mean±standard deviation. If the quantitative data were normally distributed, an independent t-test was used for analysis. The quantitative data of non-normal distribution were analyzed by nonparametric test. The counting data were compared by χ² or Fisher’s exact test. Logistic regression analysis was used in univariate analysis and multivariate analysis. Receiver operator characteristics (ROC) curve analysis was performed to assess the ultrasound diagnostic efficiency. The area under the ROC curve (AUC) was compared using the Z test. Weighted kappa (k) statistics evaluated the agreement of VTI elastography among observers. The agreement was rated as poor (k < 0.2), moderate (k = 0.2–0.4), fair (k = 0.4–0.6), good (k = 0.6–0.8), and almost perfect (k = 0.8–1.0). The difference was considered to be statistically significant when P < 0.05.

3 Results

Forty-nine women with 55 lesions were included in this study, of which 19 were malignant, and 36 were benign. The pathological classification of the lesions is described in Table 1.

Table 1
The essential characteristics of the patients and the VTI features of the breast nodules

Parameters Overall lesions (n = 55) Malignant lesions (n = 19) Benign lesions (n = 36) P

Age(years) 44.69±15.47 59.05±11.39 37.11±11.47 < 0.001

Tumor size(mm) 20.94±12.18 20.02±10.30 21.43±13.17 0.688

CMV < 0.001^*

Score 1 6 0 6

Score 2 21 1 20

Score 3 18 8 10

Score 4 10 10 0

GSV < 0.001^*

Score 1 3 0 3

Score 2 13 0 13

Score 3 15 2 13

Score 4 20 14 6

Score 5 4 3 1

Score 6 0 0 0

Pathology Invasive ductal carcinoma(13) Fibroadenoma(28)

Intraductal carcinoma(1) Intraductal papilloma(2)

Invasive lobular carcinoma(5) Benign leaf tumor(1)

Adenopathy and duct expansion(5)

Parameters	Overall lesions (n = 55)	Malignant lesions (n = 19)	Benign lesions (n = 36)	P
Age(years)	44.69±15.47	59.05±11.39	37.11±11.47	< 0.001
Tumor size(mm)	20.94±12.18	20.02±10.30	21.43±13.17	0.688
CMV				< 0.001^*
Score 1	6	0	6
Score 2	21	1	20
Score 3	18	8	10
Score 4	10	10	0
GSV				< 0.001^*
Score 1	3	0	3
Score 2	13	0	13
Score 3	15	2	13
Score 4	20	14	6
Score 5	4	3	1
Score 6	0	0	0
Pathology		Invasive ductal carcinoma(13)	Fibroadenoma(28)
		Intraductal carcinoma(1)	Intraductal papilloma(2)
		Invasive lobular carcinoma(5)	Benign leaf tumor(1)
			Adenopathy and duct expansion(5)

VTI = virtual touch tissue imaging; CMV = color-map VTI; GSV = gray-scale VTI. ^*Fisher’s exact test.

The age of patients with benign lesions was lower than that of patients with malignant lesions. (37.11±11.47 years vs. 59.05±11.39 years, P < 0.001). Most of the lesions with CMV score > 2 or GSV score > 3 were malignant (P < 0.001) (Table 1).

Univariate analysis showed that older age and higher CMV scores were significantly associated with the risk of malignancy. Multivariate analysis using logistic regression models further showed that patients with higher CMV scores (P = 0.014, odds ratio [OR] = 13.667, 95% confidence interval [CI] = 1.702–109.773) were independent predictors of breast cancer (Table 2).

Table 2

Univariate and multivariate analysis of predictive factors of malignant nodules in patients with breast lesions

	Univariate analysis			Multivariate analysis
	P	OR	95% CI	P	OR	95% CI
Age(years)	< 0.001	1.169	1.076–1.270	0.054	1.083	0.999–1.174
Tumor size(mm)	0.682	0.990	0.944–1.038	0.854	0.994	0.927–1.065
CMV	< 0.001	29.742	4.044–218.771	0.014	13.667	1.702–109.773

95% CI = 95% confidence interval; OR = odds ratio; CMV = color-map VTI.

The best cut-off value of the CMV score is > 2, CMV with a sensitivity of 94.47 (95% CI: 74.0% –99.9%) (18/19), a specificity of 72.22% (95% CI: 54.8% –85.8%) (26/36), and an AUC value of 0.912 (95% CI:0.805–0.972). The sensitivity and AUC value of CMV were higher than those of GSV, but the difference was not significant (P = 1.000 and 0.4619, respectively) (Table 3) (Fig. 3).

Table 3

Diagnostic performances of CMV and GSV

Data^a	Cut-off value	Sensitivity(%)	Specificity(%)	AUC
CMV	> 2	94.74 (74.0–99.9) [18/19]	72.22 (54.8–85.8) [26/36]	0.912 (0.805–0.972)
GSV	> 3	89.47 (66.9–98.7) [17/19]	80.56 (64.0–91.8) [29/36]	0.876 (0.760–0.950)
χ² or Z		0.314	0.693	0.736
P ^b		1.000^c	0.405	0.4619

AUC = area under ROC curve; CMV = color-map VTI; GSV = gray-scale VTI. ^a = Data are percentages, with 95% confidence intervals in parentheses and raw data in brackets. ^b = CMV compare with GSV. ^c = Fisher’s exact test.

Fig. 3

Receiver operating characteristic curve analysis of CMV and GSV for discrimination between benign and malignant breast lesions.

The agreement between observer A and observer B of the CMV score was almost perfect, with a k value of 0.854. The inter-observer agreement of the GSV scores was fair (k = 0.599) (Table 4).

Table 4

Inter-observer agreement for VTI score.

	Weighted kappa	SE	95% CI
CMV	0.854	0.049	0.758–0.950
GSV	0.599	0.077	0.448–0.750

CMV = color-map VTI; GSV = gray-scale VTI; SE = standard error; 95% CI = 95% confidence interval.

4 Discussion

VTI is a gray-scale or color-coded map that reflects the relative hardness of the tissue. Two interpretation methods have been proposed in VTI evaluation: the ratio of area size to US and VTI classification. Studies have shown that the VTI area of malignant lesions appears to be larger than that of the US, while benign lesions are similar. According to the degree of gray-scale lesions, VTI classification’s relative hardness standard has been proposed and used to diagnose thyroid and breast lesions. In this study, hard lesions indicated by VTI scores were associated with malignant lesions, consistent with previous studies [14 , 19].

Our results showed that most of the CMV scores of malignant lesions were 3 and 4, and the CMV scores of benign lesions were 1 and 2 (P < 0.001). In this study, the sensitivity, specificity, and AUC values of CMV were 94.74%, 72.22%, and 0.912, respectively, similar to those of Zhou et al. However, the diagnostic cut-off value of Zhou et al. was > 3, while ours was > 2, which might be related to the sample size [18].

Our results showed that the VTI score was 4, indicating that the whole lesion and its surrounding tissue were highly stiff, mainly in malignant breast lesions. Pathologically, some breast cancer can infiltrate the adjacent breast tissue and increase the surrounding tissue’s hardness, which may not be seen in the US. Still, the enlargement of the VTI sclerosing area can be seen on VTI. Therefore, the VTI size is larger than that of the US [6, 16].

In this study, when CMV was used to diagnose breast lesions, there were one false negative and ten false positives, which may be due to the hardness overlap between benign and malignant breast lesions. Some malignant lesions may be soft if combined with bleeding or liquefaction, and some mature fibroadenomas may be very stiff, which may interfere with the diagnosis of CMV. MRI or contrast-enhanced ultrasound (CEUS) may be used as a supplementary diagnostic method to provide more diagnostic information.

In this study, compared with GSV, the sensitivity and AUC value of CMV were higher, although the difference was not significant. However, CMV uses five colors to represent different hardness, which is more practical and comfortable operating in clinical work. Our results confirmed that CMV was a highly repetitive technique with almost perfect consistency (0.854) among radiologists with different work experience. At the same time, the k value of agreement among GSV observers was only 0.599. Therefore, CMV is a useful method for the diagnosis of breast lesions.

This study has several limitations. First of all, this study is a retrospective analysis, in which the selection bias is inevitable. Also, only pathologically confirmed nodules are included, the sample size is not large enough, and the pathological type is relatively simple. Second, the radiologists involved in image interpretation in this study come from the same institution and the same department. Observers who have been trained and educated in different institutions may lead to different results. Third, in this study, the images analyzed by the observers are static, not real-time. During the real-time inspection, diagnostic performance and agreement among observers may differ, representing day-to-day practice. Fourth, this study did not compare the diagnostic performance of CMV with MRI or CEUS. Fifth, the size of the VTI’s ROI is critical. If the lesion is close to the skin or chest, high-quality VTI images may not be obtained, and the results may be unreliable. Finally, VTI’s role in nodules with calcifications may be limited because calcification can lead to an increase in hardness.

5 Conclusion

To sum up, this study shows that CMV is a useful tool for differentiating benign and malignant breast lesions. Compared with GSV, the inter-observer agreement of CMV is higher.

Footnotes

Acknowledgments

The authors gratefully acknowledge the technical assistance of Xiao-Jia Liu, Xiao-E Huang, Xiaohuang-Zhu, and Wenqiang-Lin from the Department of Ultrasound, The First Affiliated Hospital of Shantou University Medical College, Shantou, Guangdong Province, P.R. China.

Conflict of interest

The authors declare that they have no conflict of interest.

Ethical approval

Our institutional ethics committee has approved this retrospective study. The Institutional Review Board waived informed consent.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

References

Joe

. The evolution of breast imaging: past to present. 2014;273:S23–44.

Liang , Ou , Wu , et al. Combined use of strain elastography and superb microvascular imaging with gray-scale ultrasound according to the BI-RADS classification for differentiating benign from malignant solid breast masses. Clin Hemorheol Microcirc. 2020;74:391–403.

Georgieva , Prantl , Utpatel , et al. Diagnostic performance of ultrasound strain elastography for differentiation of malignant breast lesions. Clin Hemorheol Microcirc. 2019;71:237–47.

Gundry, gynecology. Breast Ultrasound: Indications and Findings. 2016;59:380–93.

Xue

, Li

. Value of shear wave elastography in discriminating malignant and benign breast lesions: A meta-analysis. 2017;96:e7412.

Barr

. The Role of Sonoelastography in Breast Lesions. 2018;39:98–105.

Luo , Cao , Nian , et al. Benefit of Shear-wave Elastography in the differential diagnosis of breast lesion: a diagnostic meta-analysis. 2018;1:43–9.

Faruk

, Arefin

. The Journey of Elastography: Background, Current Status, and Future Possibilities in Breast Cancer Diagnosis. 2015;15:313–24.

Carlsen , Ewertsen , Sletting , et al. Ultrasound Elastography in Breast Cancer Diagnosis. 2015;36:550–62; quiz 63-5.

10.

Gkali , Chalazonitis , Feida , et al. Breast Elastography: How We Do It. 2015;31:255–61.

11.

Liu , Zheng , Shan , et al. elastography by acoustic radiation force impulse technology for differentiation of benign and malignant breast lesions: a meta-analysis. 2016;43:47–55.

12.

Zhang , Wan , Liu , et al. Value of conventional ultrasound, ultrasound elasticity imaging, and acoustic radiation force impulse elastography for prediction of malignancy in breast lesions. Clin Hemorheol Microcirc. 2020;74:241–53.

13.

Li , Xu , Bo , et al. Value of Virtual Touch Tissue Imaging Quantification for Evaluation of Ultrasound Breast Imaging-Reporting and Data System Category 4 Lesions. Ultrasound Med Biol. 2016;42:2050–7.

14.

Pu , Zhao , Yao , et al. Conventional US combined with acoustic radiation force impulse (ARFI) elastography for prediction of triple-negative breast cancer and the risk of lymphatic metastasis. Clin Hemorheol Microcirc. 2017;65:335–47.

15.

Wan , Wu , Yao , et al. Acoustic radiation force impulse elastography in evaluation of triple-negative breast cancer: A preliminary experience. Clin Hemorheol Microcirc. 2018;70:301–10.

16.

Wang

, Du

. Preliminary Results of Acoustic Radiation Force Impulse Imaging by Combined Qualitative and Quantitative Analyses for Evaluation of Breast Lesions. J Ultrasound Med. 2018;37:2405–12.

17.

Zhao

, Sun

, Tao

. Comparison of strain and acoustic radiation force impulse elastography of breast lesions by qualitative evaluation. Clin Hemorheol Microcirc. 2018;70:39–50.

18.

Zhou , Yang , Zhan , et al. Breast Lesions Evaluated by Color-Coded Acoustic Radiation Force Impulse (ARFI) Imaging. 2016;42:1464–72.

19.