Objectives— The purpose of this study was to investigate the clinical usage of Virtual Touch tissue quantification (VTQ; Siemens Medical Solutions, Mountain View, CA) implementing sonographic acoustic radiation force impulse technology for differentiation between benign and malignant solid breast masses.
Methods— A total of 143 solid breast masses were examined with VTQ, and their shear wave velocities (SWVs) were measured. From all of the masses, 30 were examined by two independent operators to evaluate the reproducibility of the results of VTQ measurement. All masses were later surgically resected, and the histologic results were correlated with the SWV results. A receiver operating characteristic curve was calculated to assess the diagnostic performance of VTQ.
Results— A total of 102 benign lesions and 41 carcinomas were diagnosed on the basis of histologic examination. The VTQ measurements performed by the two independent operators yielded a correlation coefficient of 0.885. Applying a cutoff point of 3.065 m/s, a significant difference (P < .001) was found between the SWVs of the benign (mean ± SD, 2.25 ± 0.59 m/s) and malignant (5.96 ± 2.96 m/s) masses. The sensitivity, specificity, and area under the receiver operating characteristic curve for the differentiation were 75.6%, 95.1%, and 85.6%, respectively. When the repeated non-numeric result X.XX of the SWV measurements was designated as an indicator of malignancy, the sensitivity, specificity, and accuracy were 63.4%, 100%, and 89.5%.
Conclusions— Virtual Touch tissue quantification can yield reproducible and quantitative diagnostic information on solid breast masses and serve as an effective diagnostic tool for differentiation between benign and malignant solid masses.
Breast cancer is a serious health threat worldwide and also the number one killer of women in China. For successful management of breast cancer, early detection is the key. Sonography has a long-established role in the assessment of mammographic and palpable abnormalities in the breast. It has been proven useful to differentiate benign and malignant solid masses.
In addition to conventional sonography, elastography is presently used to aid the differential diagnosis because it can yield information not only on the morphologic characteristics but also on the tissue elasticity of the masses. However, this technique has its own limitations: it is a qualitative evaluation method in which the acquisition of strain images requires external compression.
Now, a new trend has arisen, which applies acoustic radiation force impulse (ARFI) imaging to elastography. This technique requires no external compression and exploits short-duration acoustic radiation forces to generate localized tissue displacements. Such displacements can be tracked by sonographic correlation-based means and are related to the viscoelastic properties of local soft tissue. As such, ARFI imaging can enable qualitative visual and quantitative value measurements, and it has so far been used to delineate deep tissue structures via its viscoelastic characteristics in numerous applications.
Recently, Virtual Touch tissue quantification (VTQ; Siemens Medical Solutions, Mountain View, CA), which uses ARFI technology, has become available for diagnosis of superficial tissue lesions. However, to our knowledge, the diagnostic performance of this approach in solid breast masses has not yet been evaluated. The purpose of this study was to investigate the clinical use of VTQ for differentiation between benign and malignant solid breast masses.
Materials and Methods
The study was approved by the Institutional Review Board and Ethics Committee of Shanghai First People’s Hospital, and all participants signed informed consent forms before the study started. From January 2011 to May 2011, a total of 108 women (age range, 19–87 years; mean age, 44 years) participated in the study. All of the patients were recruited on the basis that they had been suspected to have solid breast masses based on conventional sonographic examinations. Among all of the patients, 83 had solitary masses, and 25 had multiple masses. When multiple masses were found, masses larger than 0.5 × 0.6 cm were evaluated. Hence, a total of 143 masses in the 108 patients constituted the sample.
Conventional Sonography and VTQ
All sonographic examinations were performed by one of two radiologists, each of whom had no less than 11 years of experience in breast sonography and were also well trained in VTQ. Examinations of 30 masses were performed by both radiologists independently to evaluate the reproducibility of the VTQ results. Both the conventional sonographic and VTQ measurements were performed with an Acuson S2000 ultrasound system (Siemens Medical Solutions) using a linear 9-MHz multifrequency transducer.
Virtual Touch tissue quantification tracks a shear wave in the region of interest that travels perpendicular to the direction of the acoustic push pulse and calculates the speed of the wave=10 m/s. The stiffer the tissue is, the greater the shear wave velocity (SWV) will be. In this way, VTQ can provide numeric values offering quantitative information on the tissue stiffness at a precise image-based anatomic location. At present, VTQ can be integrated into the Acuson S2000 system and performed with a conventional 9-MHz superficial transducer during a routine sonographic examination without any special preparations.
During the study, conventional sonography was performed to scan the patient’s breast thoroughly before VTQ measurement. The maximum diameters of individual masses were measured. Planes of the maximum diameters were selected for VTQ measurement.
For VTQ measurement, the patient needed to lie in a position identical to the one for the conventional sonography examination. The transducer was gently applied together with a sufficient amount of contact gel to avoid generation of artifactual areas of stiffness radiating from the skin surface. After activating VTQ, the transducer was kept still, and the patient was asked to hold her breath during acquisition of the SWV.
Region of interest completely within the mass and without thick calcifications.
The VTQ region of interest was determined to be a rectangle with fixed dimensions of 0.5 × 0.6 cm. Three regions of interest were localized to 3 areas of the selected plane to evaluate the average rigidity of the whole mass. Each region of interest was placed completely within the mass and included no thick calcifications (Figure 1). The reference region of interest was placed in the normal breast tissue at the same depth and no less than 0.5 cm away from the mass. For each region of interest, the SWV was measured at least 3 times to acquire 3 valid values, and only consistently stable values were used in the analysis. Hence, 9 SWV values were obtained from an individual mass and 3 from the reference breast tissue. All of the SWV values for an individual mass or the reference breast tissue were averaged to produce a mean SWV.
Data were expressed as mean ± standard deviation. A correlation coefficient was calculated by bivariate correlation analysis. The size of a mass, its SWV, and the SWV of the reference breast tissue were compared between the benign and malignant groups by the Mann–Whitney U test. The SWVs of the masses and reference breast tissue were compared by a paired samples t test. A receiver operating characteristic curve was calculated to assess the clinical usefulness of the SWVs. All analyses were performed with SPSS version 11.0 software for Windows (SPSS Inc, Chicago, IL), and 2-sided P < .05 was considered statistically significant.
A total of 108 women with 143 solid breast masses constituted the study group. The maximum diameters of the masses ranged from 0.8 to 4.1 cm (mean ± SD, 1.84 ± 0.63 cm). After resection, all masses were proven to be solid. Histopathologic analysis revealed 102 benign breast lesions and 41 breast carcinomas. Benign lesions consisted of fibroadenomas (n = 85), intraductal papillomas (n = 5), and adenosis (n = 12), and malignant lesions included invasive ductal carcinomas (n = 34), ductal carcinomas in situ (n = 5), a neuroendocrine carcinoma (n = 1), and a basal-like carcinoma (n = 1).
The correlation between the VTQ-measured SWV results acquired by the two independent operators is shown in Figure 2. Its correlation coefficient was 0.885, indicating that these SWV measurements were reproducible.
The maximum diameters of the benign masses ranged from 0.8 to 3.5 cm (mean, 1.77 ± 0.56 cm). Numeric SWV values could be measured for all 102 masses and the reference breast tissue. The SWVs of the benign masses ranged from 1.27 to 4.88 m/s (mean, 2.25 ± 0.59 m/s), whereas those of the reference breast tissue ranged from 0.87 to 2.62 m/s (mean, 1.54 ± 0.36 m/s). There was a significant difference in the SWVs between the benign masses and reference breast tissue (P < .001).
The maximum diameters of the malignant masses ranged from 0.8 to 4.1 cm (mean, 2.01 ± 0.75 cm). For 15 breast carcinomas (36.6%; 4 ductal carcinomas in situ and 11 invasive ductal carcinomas), numeric SWV values could be measured whereas for the remaining 26 (63.4%; 1 ductal carcinoma in situ, 1 neuroendocrine carcinoma, 1 basal-like carcinoma, and 23 invasive ductal carcinomas), all or part of the SWV measurements produced non-numeric results, which were all expressed as X.XX. It was known that the SWV values set by the system, representing the solid biological tissue values of all measurements, should be within the range of 0 to 9.10 m/s. In our study, X.XX measured in the solid target using a rigorous method was replaced by a value of 9.10, and in this way, the SWVs of the malignant masses were represented by a range of values from 1.17 to 9.10 m/s (mean, 5.96 ± 2.96 m/s).
Contrarily, all SWVs of the reference breast tissue were numeric, ranging from 0.81 to 2.95 m/s (mean, 1.68 ± 0.54m/s). There was a significant difference in the SWVs between the malignant masses and reference breast tissue (P< .001).
Benign and Malignant Group Comparison
A comparison of the sizes and SWVs between the benign and malignant groups is shown in Table 1. The SWVs of the malignant masses were significantly faster than those of the benign masses (P< .001; Figure 3).
Correlation between the shear wave velocity (SWV) results acquired by the two independent operators from the first 30 masses (r = 0.885).
Comparison of the Sizes and Shear Wave Velocities in the Benign and Malignant Groups
When the masses with all numeric SWV values were designated as benign lesions and those with non-numeric SWV values, totally or partially, expressed as X.XX, were designated as malignant lesions, the sensitivity of the differentiation between the benign and malignant lesions was 63.4%; specificity, 100%; positive predictive value, 100%; negative predictive value, 87.2%; and accuracy, 89.5%.
Receiver operating characteristic curve analysis of SWVs for differentiation between the malignant and benign solid breast masses gave a cutoff value of 3.065 m/s. When the masses whose SWVs were less than 3.065 m/s were designated as benign lesions and the masses with SWVs of greater than 3.065 m/s were designated as malignant lesions, the sensitivity of the differentiation reached 75.6%, and the specificity, positive predictive value, area under the receiver operating characteristic curve, negative predictive value, and accuracy were 95.1%, 85.6%, 86.1%, 90.7%, and 89.5%, respectively (Figures 4 and 5).
Shear wave velocity (SWV) values for the benign and malignant masses. The boxes indicate the values from the lower to the upper quartiles (25th–75th percentiles); center lines, medians; whiskers, minimum to maximum values; and dot, extreme value. The square is an outlier.
Among the 5 benign lesions whose SWVs were above the 3.065-m/s threshold, 2 fibroadenomas, 2 intraductal papillomas, and 1 adenosis were found, and for the 10 malignant lesions with SWVs below the 3.065-m/s threshold, 4 ductal carcinomas in situ and 6 invasive ductal carcinomas were found.
Receiver-operator characteristic curve of the shear wave velocity (SWV) that distinguishes malignant solid breast masses. The sensitivity, specificity, and area under the curve were 75.6%, 95.1%, and 85.6%, respectively, when a cutoff value of 3.065 m/s was applied.
In recent years, there has been increasing interest in assessing tissue elastic properties by sonography. Elastography of soft tissue relies on the deformation generated by an imparted force on the target organ.This method is based on two major imaging techniques. The first is strain elasticity imaging, also called static elastography, including real-time tissue elastography (Hitachi Medical Systems, Tokyo, Japan), eSie Touch (Siemens Medical Solutions), and elasticity imaging, among others; its implementation requires continuous transducer compression or external mechanical compression through the respiratory movements and cardiac pulsations. Its main drawback is that the compression cannot be quantified, and the site of compression cannot be restricted to the specific areas under investigation, leading to movement of the target and distortion of the measured results. The second type is acoustic stress elasticity imaging or dynamic elastography, including supersonic shear imaging and ARFI imaging, which applies a short-duration acoustic radiation force to the region of interest without producing movement of the whole target. Moreover, the acoustic radiation force can be quantified and yield quantitative information. This technique makes measured results less reliant on operator maneuvers. The advantages of this technique have been documented in other studies as well as ours.
Breast static elastography has diagnostic performance similar to that of conventional sonography for differentiating benign and malignant breast masses, but its reliability can be hampered by interobserver variability.Because the observational indicators in our study were numeric values, such variability was unlikely to arise.
Distribution of shear wave velocity (SWV) values of benign and malignant masses. The dotted line represents the cutoff value of 3.065 m/s.
In our study, for 63.4% (26 of 41) of the breast carcinomas, all or part of the SWV measurements were nonnumeric values, expressed as X.XX. There could be two main reasons to account for the X.XX values: First, the method does not conform to the biomechanical testing standard, or shear waves cannot be generated and propagated in the target; ie, the signal does not meet the quality assurance setup in the system because of considerable movement of the target (eg, due to respiration) during sampling, rendering the results unreliable, or the target is simple fluid in which shear waves could not be generated and propagated. Second, the target is so hard that the results are beyond the solid biological tissue value of 9.10 m/s set by the system. Because of the fact that we adopted a rigorous method in our study (eg, during sampling, the probe was kept still, and the patients were required to hold their breath to attain reliable SWV measurements) and verified the solid masses by histologic examination, the first reason can well be excluded. The X.XX values were most likely due to the presence of abnormal tissue (eg, dense fibrous desmoplastic tissue); for this reason, our substitution of X.XX with a value of 9.10 m/s could be justified. This argument can be strengthened by the fact that the X.XX values did not appear in any benign masses (in the absence of any thick calcification) but were only shown in the malignant lesions, with specificity and a positive predictive value of 100% (26 of 26). Such results indicate the potential clinical value of VTQ in differentiating malignant masses.
The results of this study showed that the SWVs of the benign masses were significantly faster than those of the reference breast tissue but slower than those of the malignant masses, implying that benign masses tend to be harder than normal breast tissue but softer than malignant masses, a result consistent with all previous findings. The malignant masses usually were very firm because of dense fibrous desmoplastic tissue. Our results have illustrated the clinical feasibility of using VTQ to quantitatively assess the relative stiffness of breast tissue. Furthermore, we have shown that SWV is indeed useful for differentiation between benign and malignant breast masses; in our study, an SWV value of greater than 3.065 m/s was indicative of malignancy. This finding not with standing, there was overlap. Five benign masses were recognized as malignant because their SWVs were above the 3.065-m/s threshold. Histopathologic examination showed that these 5 masses contained plentiful fibrous tissue, which might have contributed to the higher SWVs. On the other hand, 10 malignant masses were misdiagnosed as benign because their SWVs were below the 3.065-m/s threshold. Histopathologic examination revealed that these 10 masses were rich in epithelial cells and short of fibrous tissue, which might have accounted for the lower SWVs. It is also worthy of note that 80% (4 of 5) of the ductal carcinomas in situ had SWVs of less than 3.065 m/s, indicating that for detection of this specific type of malignancy, VTQ may be a less optimal approach.
The major limitations of applying VTQ in our study were the fixed box dimension of the target region of interest and its sensitivity to movement artifacts. In addition, our study was based on a relatively small number of malignant cases and types; to further confirm the benefits of VTQ, a larger sample size may be needed.
In conclusion, VTQ can yield quantitative information on the tissue stiffness of solid breast masses in a reproducible way. With a cutoff SWV value of 3.065 m/s, we could accurately identify both benign and malignant solid masses. Moreover, all of the lesions with an SWV of X.XX were found to be malignant, suggesting that repeated X.XX values can serve as a malignancy indicator. Given these promising results, we propose the use of VTQ as an effective diagnostic tool for differentiating between benign and malignant solid breast masses.
© 2012 by the American Institute of Ultrasound in Medicine
Trong nghiên cứu của chúng tôi, có 63.4% (26/41) carcinomas vú, tất cả hay một phần của đo SWV là các giá trị không số (nonnumeric), biểu hiện như X.XX. Có thể có hai lý do chính cho các giá trị X.XX: trước tiên, phương pháp thử nghiệm chuẩn biomechanical không phù hợp, hoặc sóng biến dạng không thể được tạo ra và truyền trong mục tiêu; tức là, các tín hiệu không đáp ứng các thiết lập bảo đảm chất lượng trong máy vì mục tiêu không chuyển động đáng kể (ví dụ như, nhờ sự hô hấp) trong thời gian lấy mẫu, kết xuất các kết quả không đáng tin cậy, hoặc mục tiêu là dịch đơn thuần nên sóng biến dạng có thể không được tạo ra và lan truyền. Thứ hai, mục tiêu quá cứng nên kết quả vượt quá giá trị rắn sinh học mô 9,10 m/s của máy. Vì thực tế chúng tôi đã thông qua một phương pháp nghiêm ngặt trong nghiên cứu (ví dụ như, trong khi lấy mẫu, đầu dò được giữ yên, và bệnh nhân được yêu cầu nín thở để có kết quả đo đạc SWV đáng tin cậy) và xét nghiệm mô bệnh học khối cứng, nên nguyên nhân đầu tiên cũng có thể được loại trừ. Các giá trị X.XX rất có thể do có mô bất thường (ví dụ như, nhiều mô sợi [desmoplastic]); vì lý do này, chúng tôi thay thế X.XX bằng giá trị 9,10 m/s có thể được chứng minh. Điều suy luận này được tăng cường bởi thực tế là giá trị X.XX đã không xuất hiện trong bất kỳ khối u lành tính nào (không có vôi hoá dày) mà chỉ xuất lộ trong các tổn thương ác tính, với độ đặc hiệu và giá trị tiên đoán dương là 100% (26 / 26 ca). Kết quả như vậy cho thấy giá trị tiềm năng lâm sàng của VTQ trong phân biệt các u ác tính.
Kết quả của nghiên cứu này cho thấy rằng SWVs của u lành tính nhanh đáng kể hơn mô vú tham chiếu nhưng chậm hơn so với u ác tính, ngụ ý rằng u lành tính có xu hướng cứng hơn mô vú bình thường nhưng mềm hơn u ác tính, phù hợp với tất cả phát hiện trước đây. Bởi u ác tính thường rất chắc vì dày đặc mô xơ hoá. Kết quả của chúng tôi đã minh họa tính khả thi lâm sàng của sử dụng VTQ để đánh giá định lượng độ cứng tương đối của mô vú. Hơn nữa, chúng tôi đã chứng tỏ SWV thực sự giúp ích phân biệt giữa u vú lành tính và ác tính; trong nghiên cứu của chúng tôi, giá trị SWV lớn hơn 3.065 m/s được coi là ác tính. Dấu hiệu này chưa vững vì đã có trùng lặp (overlapping). Năm khối lành tính được nhận là ác tính bởi vì SWVs đã ở trên ngưỡng 3,065m/s. Kết quả mô bệnh học cho thấy 5 u này có nhiều mô xơ, có thể đã góp phần làm cho SWVs cao hơn. Mặt khác, 10 khối ác tính đã lầm là lành tính bởi vì SWVs dưới ngưỡng 3,065m/s. Kết quả mô bệnh học cho thấy 10 khối này có nhiều tế bào biểu mô và ít mô xơ, làm cho SWVs thấp hơn. Đáng lưu ý rằng 80% (4 / 5) của ductal carcinomas in situ có SWVs thấp hơn 3,065 m/s, chỉ ra rằng với đặc trưng của loại bệnh lý ác tính này, VTQ là cách tiếp cận ít tối ưu.
Những hạn chế chủ yếu của việc áp dụng VTQ trong nghiên cứu của chúng tôi là kích thước cố định của ROI box và độ nhạy với artifact do chuyển động. Ngoài ra, nghiên cứu của chúng tôi chỉ có một số trường hợp ác tính tương đối ít và vài loại; nên để xác định lợi thế của VTQ, cần một kích thước mẫu lớn hơn.
Tóm lại, VTQ cho thông tin định lượng về độ cứng khối vú đặc có tính lập lại. Với một giá trị SWV =3,065 m/s, chúng tôi có thể xác định chính xác các u đặc cả lành tính và ác tính. Hơn nữa, với tất cả tổn thương ác tính có SWV =X.XXm/s, cho thấy rằng các giá trị X.XX lặp đi lặp lại có thể được xem như là chỉ báo ác tính. Với kết quả đầy hứa hẹn, chúng tôi đề xuất việc sử dụng VTQ như một phương tiện chẩn đoán hiệu quả cho việc phân biệt u vú đặc lành tính và ác tính.