Width and slope of the inclusion interfaces in the simulated and the experimental phantom images. “Interface 1” and “Interface 2” (vertical straight lines in Fig. 5) correspond to the upper and lower regions, respectively, between the inclusion and the surrounding medium. Best results ARN-509 shown in bold for each case.ImagesSimulationExperimentationLRHRLRHRInterface 1 width (mm)1.480.761.980.53Interface 2 width (mm)1.471.101.932.14Interface 1 mean slope3.087.272.469.91Interface 2 mean slope2.643.792.473.42Full-size tableTable optionsView in workspaceDownload as CSV
4.2. In-vivo thyroid images
The proposed algorithm was also used to process 15 US frames of an in vivo thyroid with a malignant tumour. The images were initially acquired for US static elastography, while applying a small compression with the US probe. The average displacements between two consecutive frames are 0.031 mm and 0.046 mm in the axial and lateral directions, respectively. More details about the thyroid data may be found in . The results are displayed in Fig. 8. Though the evaluation of in vivo data is more complicated, results show a better accuracy of the thyroid overall structure according to the delineation provided by a specialist doctor and shown in Fig. 8(a). Though confirmed by elastograms in , this manual delineation might not be entirely accurate and is hence used here for indication only.