Hybrid Algorithm for Elastography to Visualize Both Solid and Fluid-Filled Lesions

We propose a novel strain estimation technique that can produce reliable strain images of both solid and fluid-filled lesions. In our method, a kernel-based correlation coefficient technique and a speckle tracking-based strain estimation technique are combined into a single algorithm. The elegance of our algorithm is that fluid-filled lesions are first automatically identified by three selection criteria, and strain in those parts is estimated using the kernel-based correlation coefficient technique. Strain where fluid-filled lesions have not been detected is estimated using a speckle tracking-based algorithm, and then these two estimates are merged to form the final image. Any speckle tracking algorithm can be used in our proposed technique; however, we used a modified version of the direct average spectral strain estimation technique to describe our algorithm. We modified the direct average spectral strain estimation algorithm to track smaller strain variation and to facilitate strain calculation from multiple frames. We describe the performance of our proposed hybrid algorithm using in vivo patient data. Both the solid and fluid-filled lesions are clearly visible in the strain images produced by our proposed approach and are of better quality in terms of contrast-to-noise ratio and border sharpness than the strain images generated by other reported techniques. We also validate the performance of our proposed multiframe technique using experimental phantom data and in vivo patient data. The results reveal that the quality of the strain image can be improved using the multiframe technique compared with its dual-frame counterpart.