Quality measure boosts shear-wave elastography performance

NEW YORK CITY - The sensitivity of shear-wave breast elastography can be significantly improved by a quality measure that grades the adequacy of the shear wave itself, according to research presented at the American Institute of Ultrasound in Medicine (AIUM) annual meeting.

By incorporating a quality measure that determines whether the shear wave is sufficient in quality to yield accurate measurements, shear-wave imaging sensitivity increased from 41% to as high as 93%, according to Dr. Richard Barr, PhD, of Northeastern Ohio Medical University and Radiology Consultants in Youngstown, OH. These gains came at the cost of a nearly 17% increase in exams that couldn't be evaluated, however, Barr said.

He presented the findings during a Monday scientific session.

There are two types of elastography: strain and shear-wave imaging. Strain, or elasticity imaging (EI), is based on tissue deformation from an external or patient source and is a qualitative study. A quantitative exam, shear-wave elastography applies a "push pulse" that results in shear-wave propagation that can be measured as velocity, Barr said.

"Both strain and shear-wave imaging have been shown to improve characterization of breast masses," he said.

On strain elastography, malignant breast lesions appear larger and benign lesions appear smaller. The ratio of a length of lesion on elastography divided by lesion length on B-mode imaging can be used to characterize lesions as benign or malignant with high sensitivity and specificity, Barr said.

On shear-wave imaging, the shear-wave speed is measured on conventional B-mode images. The shear-wave velocity, Vs, is proportional to the stiffness of the tissue. The harder the lesion, the faster the Vs, he said. Vs can also be measured in a very small voxel, or point of measurement.

Some issues have been reported from shear-wave elastography, however. Some malignant lesions have not been color-coded at all on studies, and some have been coded as blue, or "soft cancer," inappropriately, Barr said.

"In our experience, approximately 50% of invasive ductal cancers either code as 'soft' or have no signal with shear-wave imaging, even if we include the 'ring' of high tissue that is often seen in some of these tissues," he said.

This problem is also not vendor-specific, having been identified on equipment from two vendors on the same case, he said.

Quality measure

In an attempt to deal with this challenge, the researchers studied the use of a quality map, which grades the shear waves themselves for their quality. The quality of the shear waves are presented in a color-coded map, with different colors used for various quality ratings.

The researchers studied the technique on 144 patients with 166 lesions who were scheduled for ultrasound-guided biopsy. Standard B-mode and color Doppler evaluation was performed, and strain imaging was acquired using a 14 MHz probe on an Acuson S2000 ultrasound scanner (Siemens Healthcare).

Shear-wave imaging was obtained using Virtual Touch Imaging Quantification (VTIQ) mode at 9 MHz on a modified S2000 system. The data was displayed as velocity map and a quality map. For strain imaging, the EI/B-mode ratio was calculated; a ratio less than 1 was considered to be a benign lesion, while a ratio of 1 or greater was considered to be malignant.

On shear-wave imaging, the highest Vs in the lesion or surrounding "ring" was documented. A note was made if there was a "ring" and the quality measurement of the ring and the central tumor was recorded. A Vs less than 4.5 m/sec was considered to be benign, while a Vs greater than 4.5 m/sec was determined to be malignant.

The 144 patients in the study had an average patient age of 48.5 (range 18-81) with an average lesion size of 10.6 mm (range 5-43 mm). The researchers correlated elastography results with pathology and obtained receiver operating characteristic (ROC) curves to assess performance.

There were 110 benign lesions and 56 malignant lesions.

Elastography performance
  Evaluable cases Sensitivity Specificity Area under the ROC curve
Strain imaging 163 95% 88% 0.9595
Shear-wave imaging without quality measure 166 41% 95% 0.6756
Shear-wave imaging with quality measure 138 78% 94% 0.7988
Shear-wave imaging with a low-quality measure and a solid lesion 166 93% 87% 0.9006

Of the benign lesions, 16 (14.6%) had no shear-wave imaging signal, 89 (81%) were benign on shear-wave imaging, and five (4.5%) were judged malignant.

In the malignant lesions, shear-wave imaging had no Vs results in 10 lesions. Twenty-five (44.6%) were considered benign and 31 (37.5%) were judged malignant.

The researchers found that the quality measure was low in all cases where no shear-wave imaging signal was obtained. It was also low in 19 (76%) of 25 soft malignant lesions and in two (40%) of five hard benign lesions. In soft benign lesions, the quality measure was low in only six (6.7%) of 89 cases.

In other findings, the researchers noted that the quality measure was high in three (12%) of 25 soft malignant lesions, which were lymphomas. Three false negatives on elasticity imaging were lymphomas, according to the researchers.

In cases where there is a low quality measure, if the lesion is solid, it is most likely a cancer, Barr said.

Both strain and shear-wave elastography provide additional information on breast lesion characterization, he concluded.

"Strain had the highest sensitivity, but shear wave had the highest specificity," Barr said. "The combination of [elasticity imaging] and [shear-wave imaging] may increase confidence and accuracy in characterizing breast masses."

Page 1 of 509
Next Page