Breast strain elastography: Observer variability in data acquisition and interpretation

Artificial intelligence-assisted ultrasound image analysis to discriminate early breast cancer in Chinese population: a retrospective, multicentre, cohort study

Background

Early diagnosis of breast cancer has always been a difficult clinical challenge. We developed a deep-learning model EDL-BC to discriminate early breast cancer with ultrasound (US) benign findings. This study aimed to investigate how the EDL-BC model could help radiologists improve the detection rate of early breast cancer while reducing misdiagnosis.

Methods

In this retrospective, multicentre cohort study, we developed an ensemble deep learning model called EDL-BC based on deep convolutional neural networks. The EDL-BC model was trained and internally validated on B-mode and color Doppler US image of 7955 lesions from 6795 patients between January 1, 2015 and December 31, 2021 in the First Affiliated Hospital of Army Medical University (SW), Chongqing, China. The model was assessed by internal and external validations, and outperformed radiologists. The model performance was validated in two independent external validation cohorts included 448 lesions from 391 patients between January 1 to December 31, 2021 in the Tangshan People's Hospital (TS), Chongqing, China, and 245 lesions from 235 patients between January 1 to December 31, 2021 in the Dazu People's Hospital (DZ), Chongqing, China. All lesions in the training and total validation cohort were US benign findings during screening and biopsy-confirmed malignant, benign, and benign with 3-year follow-up records. Six radiologists performed the clinical diagnostic performance of EDL-BC, and six radiologists independently reviewed the retrospective datasets on a web-based rating platform.

Findings

The area under the receiver operating characteristic curve (AUC) of the internal validation cohort and two independent external validation cohorts for EDL-BC was 0.950 (95% confidence interval [CI]: 0.909-0.969), 0.956 (95% [CI]: 0.939-0.971), and 0.907 (95% [CI]: 0.877-0.938), respectively. The sensitivity values were 94.4% (95% [CI]: 72.7%-99.9%), 100% (95% [CI]: 69.2%-100%), and 80% (95% [CI]: 28.4%-99.5%), respectively, at 0.76. The AUC for accurate diagnosis of EDL-BC (0.945 [95% [CI]: 0.933-0.965]) and radiologists with artificial intelligence (AI) assistance (0.899 [95% [CI]: 0.883-0.913]) was significantly higher than that of the radiologists without AI assistance (0.716 [95% [CI]: 0.693-0.738]; p < 0.0001). Furthermore, there were no significant differences between the EDL-BC model and radiologists with AI assistance (p = 0.099).

Interpretation

EDL-BC can identify subtle but informative elements on US images of breast lesions and can significantly improve radiologists' diagnostic performance for identifying patients with early breast cancer and benefiting the clinical practice.

Funding

The National Key R&D Program of China.

Different precompression does not reduce the diagnostic value of virtual touch tissue imaging and quantification (VTIQ) in breast lesions, especially for the ratio of the shear wave velocity between lesions and surrounding tissues

OBJECTIVE

To investigate the accuracy of virtual touch tissue imaging and quantification (VTIQ) in the diagnosis of benign and malignant breast lesions under four different precompression levels. The shear wave velocity (SWV) ratios of lesion to surrounding tissue were also added for diagnosis.

METHODS

167 female patients with breast lesions were included in this single center prospective study. VTIQ was performed under four different precompression levels. The SWV of the lesion, surrounding fat, and gland tissue were measured at the same depth as much as feasible 7 times. The breast lesions studied were all histopathologically confirmed. The VTIQ parameters were compared between precompression levels. The area under the receiver operating characteristic curve (AUC) was calculated to evaluate the diagnostic performance of each parameter of the VTIQ.

RESULTS

The VTIQ parameters of the malignant lesions were significantly higher than those of benign lesions in all precompression levels (P < 0.001). SWV of the lesion, fat, and gland tissue increased significantly with increasing precompression. The VTIQ parameters had great diagnostic performance for breast lesions in all precompression levels (AUC = 0.765-0.911). There was no significant difference between the precompression levels of the lesion-to-fat SWV ratio and the lesion-to-gland SWV ratio in benign and malignant lesions, and the cut-off coefficients of variation were 7.42% and 8.55%, respectively.

CONCLUSIONS

Precompression can increase the stiffness of breast lesions, fat and gland tissues, but does not reduce diagnostic value of VTIQ parameters in the breast. Under different precompression levels, the diagnosis of breast lesions by the ratio of the SWV of the lesion to the surrounding tissues is more stable.

The Role of Ultrasonography and Elastography in Differentiating Benign From Malignant Breast Masses With Pathologic Correlation

Objective:

Elastography has the potential in differentiating benign from malignant masses. The objectives of the study were to evaluate morphology of the breast masses with routine ultrasonography and elastography, to assess the role of elastography and conventional B-mode ultrasonography in differentiating benign from malignant breast masses and to correlate elastography and B-mode ultrasonography results with pathologic findings.

Materials and Methods:

This prospective observational study was conducted over a period of 18 months from January 2018 to June 2019 on 86 patients with 101 clinically palpable breast lumps who underwent B-mode ultrasonography and elastography of the breast. Baseline data, sonographic features, a modified color score, and mean strain ratio were recorded and compared with final diagnosis.

Results:

Sonography showed a sensitivity of 89.8%; specificity of 96.15%; positive predictive value (PPV) and negative predictive value (NPV) of 95.65% and 90.91%, respectively; and overall diagnostic accuracy of 93.07%. New modified dual color score showed sensitivity of 97.8%, specificity of 87.0%, PPV of 86.79%, and NPV of 87.08% with a diagnostic accuracy of 92.08%. The risk of missing a malignant case with the new modified dual color score was 2.1%. Mean strain ratio showed sensitivity of 100%; specificity of 98.11%; PPV and NPV of 97.96% and 100%, respectively; and diagnostic accuracy of 99.01%.

Conclusion:

This study demonstrates the promise of elastography in identifying possible breast malignancies, thus preventing unnecessary invasive procedures.

Interobserver and intermethod variability in data interpretation of breast strain elastography in suspicious breast lesions

Background/aim

Strain elastography has the disadvantage of being operator-dependent. Interobserver variability is observed during image acquisition and interpretation. This study aimed to analyze the interobserver and intermethod variability of strain elastography in image interpretation and evaluate the diagnostic performance combining elasticity score and strain ratio with ultrasonography.

Materials and methods

A retrospective study was conducted on 70 breast lesions evaluated with B-mode ultrasonography and strain elastography. B-mode ultrasonography findings, elasticity scores, and strain ratio values were evaluated using static images by two radiologists. BI-RADS assessment of the lesions and the decision of both observers as to whether the biopsy was required using B-mode ultrasonography, and the combined ultrasonography+elasticity score, and the combined ultrasonography+elasticity score+strain ratio were compared with the histopathological results. Also, the interobserver agreement was analyzed for all the combinations.

Results

There was very good agreement (weighted κ = 0.865) between the observers for the elasticity scores. Very good agreement was observed between the observers for BI-RADS assessments using the combined ultrasonography+elasticity score and the combined ultrasonography+elasticity score+strain ratio (weighted κ = 0.848, and 0.902, respectively). Area under the curve of B-mode ultrasonography, the combined B-mode ultrasonography+elasticity score, and the combined B-mode ultrasonography+elasticity score+strain ratio, were calculated as 0.859, 0.866, and 0.916 for observer 1, and 0.851, 0.829, and 0.916 for observer 2, respectively. There were no statistically significant differences between the observers’ diagnostic performances in any of the combinations (P = 0.703, 0.067, and 0.972, respectively).

Conclusion

In the evaluation and further assessment of breast lesions, semiquantitative strain ratio calculation may help improve diagnostic accuracy by reducing interpretational variety, when used together with B-mode ultrasonography and elasticity scoring, especially for inexperienced individuals.

Comparison of 2 shear wave elastography systems in reproducibility and accuracy using an elasticity phantom

This study aimed to compare the accuracy and inter- and intra-observer reproducibility of the measured elasticity between 2 shear wave elastography systems. Three breast radiologists examined 8 targets of 4 different levels of stiffness (size: 11 mm, 4 mm) in an elasticity phantom (Customized 049A Elasticity QA Phantom, CIRS, Norfolk, VA, USA) using 2 different shear wave ultrasound elastography systems: SuperSonic Imagine (SSI) (SSI, Aix en Provence, France) and ShearScan (RS-80A, Samsung Medison, Seoul, Korea). Three radiologists performed ultrasound (US) elastography examinations for the phantom lesions using 2 equipment over a 1-week interval. Intra- and inter-observer reproducibility and the accuracy of the measured elasticity were analyzed and compared between the 2 systems. The accuracy of shape was also analyzed by shape-matching between B-mode and elastography color image. Intra-class correlation coefficients (ICC) were used in statistical analysis. For measured elasticity, the intra-observer and inter-observer reproducibility were excellent in both SSI and ShearScan (0.994 and 0.998). The overall accuracy was excellent in both systems, but the accuracy in small lesions (4 mm target) was lower in SSI than ShearScan (0.780 vs 0.967). The accuracy of shape-matching on the elastography image was 59.0% and 81.4% in the SSI and ShearScan, respectively. In conclusion, the SSI and ShearScan showed excellent intra- and inter-observer reproducibility. The accuracy of the Young's modulus was high in both the SSI and ShearScan, but the SSI showed decreased accuracy in measurement of elasticity in small targets and poor shape-matching between the B-mode image and color-coded elastography image.

[Ultrasound elastography techniques in breast cancer]

BACKGROUND

Elasticity assessment of breast lesions can also be used as an associated criterion in the B‑mode ultrasound (US) assessment for differentiation between benign or malignant lesions.

OBJECTIVES

The goal was to identify techniques available to measure US elasticity, assess the use of B‑mode BI-RADS® classification combined with elastography, and identify which artefacts influence the US elasticity result.

MATERIALS AND METHODS

Based on different studies and meta-analyses, clinical application in daily routine of the presented US elastography techniques will be investigated concerning the statistical performance of semi-quantitative and quantitative cut-off values to differentiate benign from malignant lesions.

RESULTS

Supported by meta-analyses, all presented US elastography techniques improve the specificity by decreasing the B‑mode sensitivity. In the literature the semi-quantitative and quantitative cut-off values often vary considerably. The interobserver variability of strain elastography shows a fair agreement and the interobserver variability of shear wave elastography a substantial agreement.

CONCLUSIONS

Considering the limitations and artefacts of each technique, US elastography is able to enhance the true positive and true negative results. In the case of a higher B‑mode BI-RADS® classification (4b, 4c, 5) or in a high-risk situation to develop breast cancer, a large core needle biopsy should be performed despite lesion softness in elastography.

Similar Reproducibility for Strain and Shear Wave Elastography in Breast Mass Evaluation: A Prospective Study Using the Same Ultrasound System

The objective of this study was to evaluate the inter-operator reproducibility of strain elastography (SE) and shear wave elastography (SWE) in three groups: all lesions, benign lesions and malignant lesions. Ninety-one lesions from ninety-one women were examined by SE and SWE from January 2017 to December 2017 by two radiologists. The reproducibility of elastic score, SE strain ratio and SWE Young's modulus between operators was prospectively evaluated. There was good agreement on elasticity score, with κ values of 0.711, 0.640 and 0.766. The intra-class correlation coefficients of the strain ratio, mean elastic modulus (E_mean), maximum elastic modulus (E_max) and elastic modulus standard deviation (E_sd) ranged from 0.723-0.876, which indicated good and excellent agreement. We concluded that both SE and SWE had good reproducibility among different operators using the same probe in the same ultrasound instrument. Strain elasticity score was more consistent among operators in malignant breast tumors. There was better agreement on strain elastic ratio and shear wave elasticity among operators in benign breast lesions.

Breast sonoelastography: Now and in the future

PURPOSE

The purpose of this study was to identify practice trends and opinions concerning breast sonoelastography in two different health care systems, one in Europe (France) and the other in North America (Province of Québec/Canada).

MATERIALS AND METHODS

We distributed an 11-item online survey among French and Canadian breast radiologists. The survey comprised of four sections: (i) personal practice characteristics, (ii) breast sonoelastography usage in daily practice and evaluation of its usefulness, (iii) limitations and roles of sonoelastogrpahy in their clinical practice, and (iv) types of elastographic technique and interpretation.

RESULTS

We found that sonoelastography of the breast appears unpopular among Canadian radiologists, and poorly credible among French radiologists, who perceive it as an unreliable technique. To date, its real impact in clinical practice remains uncertain.

CONCLUSION

Continued learning and awareness of the indications, advantages and limitations of breast sonoelastography may motivate breast radiologists to adopt its use.