Quantitative analysis of elastography images in the detection of breast cancer

The aim of this study was to develop a quantitative method for breast cancer diagnosis based on elastosonography images in order to reduce whenever possible unnecessary biopsies. The proposed method was validated by correlating the results of quantitative analysis with the diagnosis assessed by histopathologic exam.

109 images of breast lesions (50 benign and 59 malignant) were acquired with the traditional B-mode technique and with elastographic modality. Images in Digital Imaging and COmmunications in Medicine format (DICOM) were exported into a software, written in Visual Basic, especially developed to perform this study. The lesion was contoured and the mean grey value and softness inside the region of interest (ROI) were calculated. The correlations between variables were investigated and receiver operating characteristic (ROC) curve analysis was performed to assess the diagnostic accuracy of the proposed method. Pathologic results were used as standard reference.

Both the mean grey value and the softness inside the ROI resulted statistically different at the t test for the two populations of lesions (i.e., benign versus malignant): p<0.0001. The area under the curve (AUC) was 0.924 (0.834-0.973) and 0.917 (0.826-0.970) for the mean grey value and for the softness respectively.

Quantitative elastosonography is a promising ultrasound technique in the detection of breast cancer but large prospective trials are necessary to determine whether quantitative analysis of images can help to overcome some pitfalls of the methodic.