Recent studies have shown that the malignancy of tumours may be inferred from their stiffness or elasticity. Benign tumours in the breast tend to have a stiffness of about 50kPa-60kPa, whilst malignant tissue tends to be stiffer at about 90kPa-120kPa. Various methods of ultrasonic elastography (elasticity imaging) exist, using methods such as speckle tracking, 2-D kernel cross-correlation, and shear wave imaging. Typically, these look at the changes between successive B-mode ultrasound images whilst the tissue is undergoing mechanical compression.
Work carried out by the Ultrasonics Research Group in UCC has investigated stiffness estimation using image segmentation and edge detection. Computer simulation was used to develop an image processing algorithm that could isolate the tumour in an image, and then monitor its change in shape using edge detction. From this, the stiffness could be estimated to within 2%-8% of the known value, depending on tumour size. The image on the right shows edge detection of the tumour at different levels of mechanical compression.
Tissue phantoms (synthetic materials that mimic the acoustic properties of real tissue) were manufactured using different mixtures of agar, gelatine, and ultrasound scattering particles to produce artificial structures with different appropriate stiffnesses. Ultrasonic B-mode images were obtained from an ESAOTE 40mm 8MHz linear array, and similar stiffness measurement accuracies were obtained. The image on the right shows a 10mm diameter region isolated at 4% compression.
- A. J. Constant and W. M. D. Wright, "Estimation of Tissue Elasticity by Image Processing of Simulated B-mode Ultrasound Images", Proc. Irish Signals and Systems Conference (ISSC 2009), Dublin, Ireland 10th-11th June 2009, in press (2009)