Impact of region of interest (ROI) size on the diagnostic performance of shear wave elastography in differentiating solid breast lesions

Abstract

Background

Shear wave elastography (SWE) using a region of interest (ROI) can demonstrate the quantitative elasticity of breast lesions.

Purpose

To prospectively evaluate the impact of two different ROI sizes on the diagnostic performance of SWE for differentiating benign and malignant breast lesions.

Material and Methods

A total of 154 breast lesions were included. Two types of ROIs were investigated: one involving an approximately 2-mm diameter, small round ROIs placed over the stiffest area of the lesion, as determined by SWE (ROI-S); and another ROI drawn along the margin of the lesion using a touch pen or track ball to encompass the entire lesion (ROI-M). Maximum elasticity (E_max), mean elasticity (E_mean), minimum elasticity (E_min), and standard deviation (SD) were measured for the two ROIs. The area under the receiver operating characteristic curve (AUC) as well as the sensitivity and specificity of each elasticity value were determined.

Results

The AUCs for ROI-S were higher than those for ROI-M when differentiating benign and malignant breast solid lesions. The E_max, E_mean, E_min, and SD of the elasticity values for ROI-S were 0.865, 0.857, 0.816, and 0.849, respectively, and for ROI-M were 0.820, 0.780, 0.724, and 0.837, respectively. However, only E_max (P = 0.0024) and E_mean (P = 0.0015) showed statistically significant differences. For ROI-S, the sensitivity and specificity of E_max were 78.8% and 84.3%, respectively, and those for E_mean were 80.8% and 81.4%, respectively.

Conclusion

Using ROI-S with E_max and E_mean has better diagnostic performance than ROI-M for differentiating between benign and malignant breast lesions.

Keywords

Breast ultrasonography elastography shear wave

Introduction

Ultrasound (US) elastography is a diagnostic technology that reflects the stiffness of a targeted lesion and many studies have shown that this technique improves the diagnostic performance of conventional US for differentiating between benign and malignant breast lesions (1 –5). As one of two types of elastography, shear wave elastography (SWE) is highly reproducible (6) and provides color elasticity maps as well as quantitative values.

SWE induces mechanical vibrations in tissue components using an acoustic radiation force generated by a focused US beam. The displacement at the focus of the beam generates a shear wave that relays information associated with the viscoelastic properties of the tissue, thereby allowing for a quantitative approach to determining elasticity values (7,8). An ultrafast US acquisition sequence is then used to capture the propagation of the shear wave (9,10). By setting the focus points within a region of interest (ROI), various quantitative elasticity values can be acquired. Based on quantitative assessments with ROIs, several studies have reported that SWE can accurately differentiate benign and malignant breast lesions (8,11 –13).

However, there currently exist no standard methods for determining ROIs. For example, some researchers have used small 1–3 mm ROIs located in the stiffest portion of the mass (9,11,14 –19), whereas others have used large ROIs that cover the entire lesion (20 –22).

To date, there have been no published evaluations comparing these two types of ROIs, namely, small ROIs located at the stiffest portion of the lesion and large ROIs encompassing the entire lesion. Therefore, the aim of the present study was to compare the diagnostic performance of these two ROI types.

Material and Methods

Patient selection and data collection

This prospectively study was approved by the Institutional Review Board of Hallym University Sacred Heart Hospital (Anyang-si, Republic of Korea). All patients provided written informed consent.

Elastography was performed in 183 patients before biopsy or surgery. We excluded patients who had undergone chemotherapy or radiotherapy for previous cancer, and patients who had undergone neoadjuvant chemotherapy before US examinations. Cases with lesions that were larger than the probe were excluded. Therefore, 152 patients with 154 breast lesions were enrolled (age range = 23–86; mean age = 56.4 years). The average diameter (±SD) of the tumor was 20.1 ± 9.8 mm (range = 50–54 mm; median = 18.0 mm). All patients with BI-RADS scores of 4 or 5 for the lesion underwent core needle biopsies (14-gauge auto- or semi-automated gun, Stericut, TSK, Laboratory, Oisterwijk, The Netherlands), and in some cases, lesions with scores of 2 or 3 were also biopsied as per the patient’s or clinician’s request. Histopathology results from biopsies and surgeries served as reference standards.

Image acquisition, conventional US

Two radiologists with five and seven years of experience, respectively, in breast US performed the entire breast US procedures in the transverse and longitudinal planes for all 152 patients as well as radial or oblique scanning of some lesions. US examinations were performed with IU-22 and EPIQ machines (Philips Healthcare, Bothwell, WA, USA) with 5–12-MHz linear transducers or an Aixplorer® system (SuperSonic Imagine, Aix en Provence, France) with a 4–15-MHz linear transducer.

The US features of each lesion were evaluated according to the fourth edition of the Breast Imaging Reporting and Data System (BI-RADS). Features were graded as follows: category 0 = assessment incomplete; category 1 = negative; category 2 = benign; category 3 = probably benign; category 4A = low suspicion for malignancy; category 4B = moderate suspicion for malignancy; category 4C = high suspicion for malignancy; category 5 = highly suggestive of malignancy, and category 6 = known biopsy-proven malignancy (23).

Image acquisition, elastography

Following conventional US, SWE was performed on targeted lesions by a single radiologist with seven years of experience in breast US. The Aixplorer® system was used and SWE images were obtained by gently applying a 4–15-MHz linear transducer to the skin above the targeted lesion with a generous quantity of gel. The probe was maintained in a still position without compression for a few seconds to capture adequate elastography images. Conventional US and SWE images were displayed in split-screen mode and the SWE image was superimposed onto the corresponding B-mode gray-scale US image. The tissue elasticity of each pixel in the SWE image was transformed into a color-coded map representing Young’s modulus (kPa), with a range from dark blue (soft) to red (stiff) (range = 0–180 kPa). The elasticity image was refreshed in real time.

Quantitative elasticity values were determined by locating the lesions in the center of an ROI box, with a default area of 2.5 × 1.5 cm and a maximal area of 3.0 × 2.5 cm (8). The radiologists placed a 2-mm round quantification ROI onto the previously generated color map. The first ROI (ROI-S) was positioned at the stiffest portion of the lesion, including the immediately adjacent stiff tissue or halo (9,14 –17). The second ROI (ROI-M) was then drawn along the margin of the lesion using a touch pen or track ball on the touch screen (Fig. 1). During this process, several values of SWE evaluation were calculated automatically and displayed on the screen. Of these values, maximum elasticity (E_max), mean elasticity (E_mean), minimum elasticity (E_min), and standard deviation (SD) for all lesions were recorded for subsequent analysis. Acquisition of the SWE images took ∼2–3 min/case.

Fig. 1.

Measurements of ROI-S and ROI-M. (a) ROI-S was a 2-mm round region positioned at the stiffest portion of the lesion. (b) ROI-M was drawn along the margin of the lesion using a freely moving touch pen or track ball to encompass the entire lesion.

Statistical analysis

The E_max, E_mean, E_min, and SD values for ROI-S and ROI-M were collected for each lesion, and the means of benign and malignant lesions were compared using a two-sample t-test. To evaluate and compare diagnostic performance, receiver operating characteristic (ROC) curves for ROI-S and ROI-M were calculated based on E_max, E_mean, E_min, and SD. Statistically significant differences between the Az values were reported with 95% confidence intervals (CI). Values of P < 0.05 were considered to indicate statistically significant differences between groups when the corresponding CIs were not zero. In addition, optimal cutoff values that yielded maximal sums of the sensitivity and specificity for ROI-S and ROI-M were calculated. The Stata software program (release version 9.0; Stata Corporation, College Station, TX, USA) was used for all statistical analyses.

Results

Of the 154 lesions, 52 (33.8%) were malignant and 102 (66.2%) were benign. The BI-RADS categories of all lesions were as follows: category 2 (benign) for two masses, category 3 (probably benign) for 42 masses, category 4a (low suspicion for malignancy) for 58 masses, category 4b (moderate suspicion for malignancy) for eight masses, category 4c (high suspicion for malignancy) for 12 masses, and category 5 (highly suggestive of malignancy) for 32 masses. The average diameter (±SD) of the benign lesions was 20.7 ± 13.8 mm (range = 4–50 mm; median = 17.3 mm), and the average diameter of the malignant lesions was 23.2 ± 17.1 mm (range = 5–55 mm; median = 16.9 mm).

Table 1 summarizes the mean values of E_max, E_mean, E_min, and SD for ROI-S and ROI-M for all lesions. The means of each elasticity value for the malignant lesions were significantly higher than those for benign lesions (P < 0.001). The areas under the curve (AUCs) for E_max, E_mean, E_min, and SD for ROI-S were 0.8650 (95% CI = 0.8019–0.9281), 0.8566 (95% CI = 0.0.7903–0.9230), 0.8164 (95% CI = 0.7334–0.8990), and 0.8494 (95% CI = 0.7908–0.9080), respectively, and for ROI-M were 0.8200 (95% CI = 0.7470–0.8931), 0.7798 (95% CI = 0.6983–0.8613), 0.7235 (95% CI = 0.6354–0.8116), and 0.8370 (95% CI = 0.7699–0.9041), respectively, for differentiating benign and malignant breast solid lesions (Table 2) (Fig. 2). All elasticity values of ROI-S were higher than those for ROI-M. However, only E_max (P = 0.0024) and E_mean (P = 0.0015) showed statistically significant differences. For ROI-S, the sensitivity and specificity of E_max were 78.8% and 84.3%, respectively, and for E_mean, these values were 80.8% and 81.4%, respectively (Table 3).

Table 1.

Mean values of E_max, E_mean, E_min, and SD for lesions using ROI-S and ROI-M.

ROI	Elasticity values	Benign	Malignancy	P value
ROI-S	E_max	71.1	190.6	0.0000
	E_mean	63.2	171.2	0.0000
	E_min	55.2	146.6	0.0000
	SD	4.9	12.6	0.0000
ROI-M	E_max	81.0	200.7	0.0000
	E_mean	36.7	84.9	0.0483
	E_min	11.3	66.9	0.0000
	SD	14.1	43.8	0.0000

ROI, region of interest; ROI-S, small ROI; ROI-M, large ROI; E_max, maximum value of elasticity; E_mean, mean value of elasticity; E_min, minimum value of elasticity; SD, standard deviation.

Table 2.

AUC values for E_max, E_mean, E_min, and SD for ROI-S and ROI-M.

ROI	Elasticity values	AUCs	95% CI
ROI-S	E_max	0.8650	0.8019–0.9281
	E_mean	0.8566	0.7903–0.9230
	E_min	0.8164	0.7334–0.8990
	SD	0.8494	0.7908–0.9080
ROI-M	E_max	0.8200	0.7470–0.8931
	E_mean	0.7798	0.6983–0.8613
	E_min	0.7235	0.6354–0.8116
	SD	0.8370	0.7699–0.9041

ROI, region of interest; ROI-S, small ROI; ROI-M, large ROI; E_max, maximum value of elasticity; E_mean, mean value of elasticity; E_min, minimum value of elasticity; SD, standard deviation.

Fig. 2.

(a–d) ROC curves for per-elasticity values of ROI-S and ROI-M: (a) S_max, maximum elasticity of ROI-S, M_max, maximum elasticity of ROI-M, (b) S_mean, mean elasticity of ROI-S, M_mean, mean elasticity of ROI-M, (c), S_min, minimum elasticity of ROI-S, M_min, minimum elasticity of ROI-M, (d) S_SD, standard deviation of ROI-S, M_SD, standard deviation of ROI-M.

Table 3.

Sensitivities and specificities of elasticity values for ROI-S and ROI-M.

Elasticity values		Sensitivity (%)	Specificity (%)
ROI-S	E_max	78.8	84.3
	E_mean	80.8	81.4
	E_min	76.9	86.3
	SD	76.9	79.4
ROI-M	E_max	69.2	89.2
	E_mean	63.4	87.3
	E_min	61.5	80.3
	SD	65.3	87.2

ROI, region of interest; ROI-S, small ROI; ROI-M, large ROI; E_max, maximum value of elasticity; E_mean, mean value of elasticity; E_min, minimum value of elasticity; SD, standard deviation.

Discussion

We found that small ROIs located at the stiffest area of the lesion had a better diagnostic performance for discriminating between benign and malignant tumors than large ROIs containing the entire lesion. However, of the several elasticity values we tested, only E_max and E_mean values corresponded to statistically significant differences between lesion types.

Some investigators have used large ROIs to cover the entire lesion, including lesion borders and any calcification, to calculate elasticity values (20,21). By contrast, other investigators have used small ROIs with diameters of approximately 1–3 mm (9,11,14 –17,19). These researchers used a variety of parameters with variable cutoff values for differentiating between benign and malignant breast lesions (Table 4), and the parameters had a range of diagnostic value. When comparing these two types of ROIs among previous studies, small ROIs appeared superior to large ROIs in some aspects. However, such comparisons are difficult to make because elastography techniques are not standardized, and investigators have a variety of ROI sizes, reference tissues, and elasticity values during examinations.

Table 4.

Comparison two types of ROIs with a variety of parameters from previous researches.

Types of ROI	Authors/years of publication/journal	Elasticitiy values (cutoff value)	Diagnostic performances (A_z value)	Sensitivities (%)	Specificities (%)
ROI-M	Wang et al./2013/Radiol Med (21)	E_max (91.5 kPa), E_mean (57.2 kPa)	0.741, 0.692	60.9, 45.7	85.6, 85.3
ROI-M	Zhou et al./2014/Radiology (20)	E_max (NM), E_mean (NM), E_min (NM), SD (NM)	0.850, 0.744, 0.623, 0.853	76.8, 55.4, 51.8, 91.1	78.1, 85.4, 86.1, 67.9
ROI-S	Au et al./2014/AJR Am J Roentgenol (16)	E_max (46.7 kPa), E_mean (42.5 kPa), E_ratio (3.6)	0.943, 0.932, 0.666	90.9, 89.9, 86.4	88.6, 89.9, 93.7
	Lee et al./2013/Eur Radiol (9)	E_max (82.3 kPa), E_mean (68.4 kPa), E_ratio (4.4), SD (9.6 kPa)	0.860, 0.844, 0.782, 0.850	88.9, 88.9, 88.9, 72.2	77.5, 76.7, 65.8, 87.5
	Olgun et al./2014/Diag Inter Radiol (14)	E_max (54.3 kPa), E_mean (45.7 kPa), E_min (37.1 kPa), E_ratio (4.7)	0.972, 0.973, 0.969, 0.987	96.4, 96.5, 96.5, 96.7	94.3, 95.3, 95.3, 96.5
	Ko et al.*/ 2014/Eur Radiol (15)	E_mean (41.6 kPa)	0.788	83.3	68.2
	Kim et al./2014/J Ultrasound Med (17)	E_max (NM), E_mean (96.8 kPa), E_ratio (NM)	0.865, 0.870, 0.849	74.6%†	87.1%‡

The target lesions are non-mass lesions.

†

The sensitivity is for E_mean.

‡

The specificity is for E_mean.

ROI, region of interest; ROI-S, small ROI; ROI-M, large ROI; E_max, maximum value of elasticity; E_mean, mean value of elasticity; E_min, minimum value of elasticity; SD, standard deviation; E_ratio, the ratio of E_mean value in the stiffest portion of the mass to the E_mean of normal fatty tissue; NM, not mentioned.

During SWE imaging, tissue elasticity is visualized as a color map overlying the gray-scale image, with red representing stiff tissue and blue representing soft tissue. In certain hard cancers, the red stiff tissue area is often located at the margin of the lesion (Fig. 3). This is known as a “stiff rim sign” in SWE and it is not observed in strain type elastography. A “stiff rim sign” means that the shear wave amplitude is low and noisy within the cancer but a smooth shear wave response can be seen in the surrounding normal tissue (24). Several theories have been proposed regarding the cause of this signal, and some investigators claim it is due to a desmoplastic reaction or the infiltration of cancer cells into interstitial tissues (19,25,26). Others have reported that this is a low shear wave amplitude and/or noise within the malignant lesion that could be caused by attenuation of the energy of the shear wave in the peritumoral region of the lesion (27). When using an ROI-M, the ROI can cover the entire lesion area; however, in the case of a stiff rim sign located outside the lesion, the stiffest site could be excluded when measuring elasticity values. By contrast, the ROI-S can be positioned at the margin or even outside the lesion, explaining the weaker diagnostic performance of ROI-M compared with ROI-S (Fig. 4). Our results indicate that it is more important to measure the stiffest area with a small ROI than to measure the stiffness of the entire lesion with a large ROI. Indeed, it was recently shown that a stiff rim sign can potentially be used to improve the differentiation of breast lesions (20).

Fig. 3.

Stiff rim sign.

Fig. 4.

Using ROI-M, the stiffest portion is located outside the lesion in some hard cancers.

Skerl et al. evaluated the influence of ROI size (1, 2, and 3 mm) and lesion diameter on the diagnostic performance of SWE in solid breast lesions. Although no single SWE parameter had superior performance in their study, they recommended an ROI of 2 mm as a good compromise.

Of the several elasticity values we tested, only E_max and E_mean showed statistically significant differences between lesion types. E_max and E_mean represent the general stiffness parameters of solid masses. However, E_min is the least reproducible metric for evaluating elasticity values (11). In this study, the diagnostic performance of E_min was the lowest for both ROI-S and ROI-M. Wang et al. (21) also reported that E_min is not useful for differentiating between benign and malignant lesions (21). SD is a measure of lesion heterogeneity, which is more marked in malignant lesions than in benign lesions (12). However, when we used ROI-S, the SD did not reflect the heterogeneity of the entire lesion. Thus, the diagnostic performance of the SD is lower than that of E_max or E_mean. However, when using ROI-M, the diagnostic performance of the SD was the best. This result indicates that the usefulness of the SD depends on ROI size and is only reliable when using ROI-M.

In this study, the sensitivity and specificity of E_max were 78.8% and 84.3%, respectively, and for E_mean 80.8% and 81.4%, respectively, for ROI-S. These results are similar to or somewhat lower than values reported in other studies (9,14,16).

There were some limitations to this prospective, single-center study. First, there could have been unavoidable selection bias because the patients included in this study were scheduled for biopsy or excision of known breast lesions. Second, the influence of lesion depth or breast thickness was not considered in this study and indeed, some studies suggest that breast thickness can influence the performance of SWE, which requires further research. Third, long-term follow-up data were not available in concordant benign lesions after biopsy. Last, we did not correlate elasticity values with histologic differentiation, histologic grade, or internal microcalcifications, any of which could influence diagnostic performance.

In conclusion, the use of ROI-S provides better diagnostic performance than ROI-M for differentiating between benign and malignant breast lesions. When using ROI-S, only E_max and E_mean showed statistically significant differences between lesion types among the several elasticity values tested.

Footnotes

Declaration of Conflicting Interests

The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) received no financial support for the research, authorship, and/or publication of this article.

References

Ianculescu

Ciolovan

Dunant

et al.

Added value of Virtual Touch IQ shear wave elastography in the ultrasound assessment of breast lesions. Eur J Radiol 2014; 83: 773–777.

Golatta

Schweitzer-Martin

Harcos

et al.

Evaluation of virtual touch tissue imaging quantification, a new shear wave velocity imaging method, for breast lesion assessment by ultrasound. Biomed Res Int 2014; 2014: 960262–960262.

Tozaki

Saito

Benson

et al.

Shear wave velocity measurements for differential diagnosis of solid breast masses: a comparison between virtual touch quantification and virtual touch IQ. Ultrasound Med Biol 2013; 39: 2233–2245.

Balleyguier

Ciolovan

Ammari

et al.

Breast elastography: the technical process and its applications. Diagn Interv Imaging 2013; 94: 503–513.

Tozaki

Fukuma

. Pattern classification of ShearWave™ Elastography images for differential diagnosis between benign and malignant solid breast masses. Acta Radiol 2011; 52: 1069–1075.

Cosgrove

Berg

Doré

et al.

Shear wave elastography for breast masses is highly reproducible. Eur Radiol 2012; 22: 1023–1032.

Ricci

Maggini

Mancuso

et al.

Elastography in the diagnosis of breast lesions: comparison of different elastographic features. Acta Radiol 2017; 58: 1189–1197.

Chang

Moon

Cho

et al.

Clinical application of shear wave elastography (SWE) in the diagnosis of benign and malignant breast diseases. Breast Cancer Res Treat 2011; 129: 89–97.

Lee

Jung

et al.

Diagnostic performances of shear wave elastography: which parameter to use in differential diagnosis of solid breast masses?

Eur Radiol 2013; 23: 1803–1811.

10.

Bercoff

Tanter

Fink

. Supersonic shear imaging: a new technique for soft tissue elasticity mapping. IEEE Trans Ultrason Ferroelectr Freq Control 2004; 51: 396–409.

11.

Berg

Cosgrove

Doré

et al.

Shear-wave Elastography Improves the Specifi city of Breast US: The BE1 Multinational Study of 939 Masses. Radiology 2012; 262: 435–449.

12.

Evans

Whelehan

Thomson

et al.

Quantitative shear wave ultrasound elastography: initial experience in solid breast masses. Breast Cancer Res Treat 2010; 12: R104–R104.

13.

Athanasiou

Tardivon

Tanter

et al.

Breast lesions: quantitative elastography with supersonic shear imaging—preliminary results. Radiology 2010; 256: 297–303.

14.

Olgun

DÇ

Korkmazer

Kılıç

et al.

Use of shear wave elastography to differentiate benign and malignant breast lesions. Diagn Interv Radiol 2014; 20: 239–244.

15.

Jung

Kim

et al.

Potential role of shear-wave ultrasound elastography for the differential diagnosis of breast non-mass lesions: preliminary report. Eur Radiol 2014; 24: 305–311.

16.

FW-F

Ghai

Moshonov

et al.

Diagnostic performance of quantitative shear wave elastography in the evaluation of solid breast masses: determination of the most discriminatory parameter. Am J Roentgenol 2014; 203: 328–336.

17.

Kim

Choi

et al.

Diagnostic performance of shear wave elastography of the breast according to scanning orientation. J Ultrasound Med 2014; 33: 1797–1804.

18.

Yoon

Kim

E-K

et al.

Interobserver variability of ultrasound elastography: how it affects the diagnosis of breast lesions. Am J Roentgenol 2011; 196: 730–736.

19.

Evans

Whelehan

Thomson

et al.

Differentiating benign from malignant solid breast masses: value of shear wave elastography according to lesion stiffness combined with greyscale ultrasound according to BI-RADS classification. Br J cancer 2012; 107: 224–229.

20.

Zhou

Zhan

Chang

et al.

Breast lesions: evaluation with shear wave elastography, with special emphasis on the “stiff rim” sign. Radiology 2014; 272: 63–72.

21.

Wang

et al.

Study of quantitative elastography with supersonic shear imaging in the diagnosis of breast tumours. Radiol Med 2013; 118: 583–590.

22.

Cho

Moon

Kim

et al.

Sonoelastographic strain index for differentiation of benign and malignant nonpalpable breast masses. J Ultrasound Med 2010; 29: 1–7.

23.

American College of Radiology. ACR BI-RADS® Atlas, Breast Imaging Reporting and Data System, Reston, VA: American College of Radiology, 2013.

24.

Barr

Zhang

. Shear wave imaging of the breast still on the learning curve. J Ultrasound Med 2012; 31: 347–350.

25.

Itoh

Ueno

Tohno

et al.

Breast disease: Clinical application of us elastography for diagnosis. Radiology 2006; 239: 341–350.

26.

Garra

Cespedes

Ophir

et al.

Elastography of breast lesions: initial clinical results. Radiology 1997; 202: 79–86.

27.

Tozaki

Isobe

Fukuma

. Preliminary study of ultrasonographic tissue quantification of the breast using the acoustic radiation force impulse (ARFI) technology. Eur J Radiol 2011; 80: 182–187.