Abstract
Nakagami parametric image has been used to discriminate different scatterer concentrations, and furthermore investigated to efficiently classify breast masses. However, the different scanning sections of breast lesions could cause misclassifications when only adopting one single Nakagami image. This study explored the feasibility of applying strain compounding technique to Nakagami image for considering multiple Nakagami images to reduce the diagnostic errors and improving the clinical performance in breast tumors classification. Strain compounding technique has been investigated to be useful for speckle reduction in conventional B-scan images due to scatterers redistribution caused by external compression. The external compression could cause tissue deformation which is relative to tissue stiffness and scatterers concentrations inside the breast tissue. Therefore, the continuous Nakagami images obtained by external compression could reveal the changes of scatterers concentration according to the stiffness of breast tissue and thus could be averaged to construct the strain-compounding Nakagami image. The difference of scatterers distribution between the original Nakagami image and the strain-compounding Nakagami image may be useful in distinguishing softer tissues from harder ones. In general, malignant breast tumor is harder than benign one due to desmoplastic reactions around surrounding tissues. Therefore, we propose the strain-compounding Nakagami imaging to classify breast tumors. The receiver operating characteristic (ROC) curve was used to evaluate the diagnostic performance. The area under the ROC curve was 0.97±0.02, the diagnostic accuracy was 96%, the sensitivity was 100%, and the specificity was 92%. In conclusion, the strain-compounding Nakagami method shows that the different degree of hardness of breast tumors can be distinguished by the changes in Nakagami parameters. Therefore, the proposed method has the potential to be a useful computer-aided diagnosis tool in the detection and characterization of the breast masses.