Attenuation Compensation and Boundary Segmentation in Ultrasound Images
Ultrasound imaging has great potential in helping physicians diagnose and treat
medical conditions. It is quick, inexpensive, boasts excellent soft tissue contrast.
The equipment is light and portable, and it produces dynamic video images for real-time
viewing of tissue motion. One important fact is that it does not expose patients to
potentially harmful ionizing radiations, such as X-rays and gamma rays. Current limitations of ultrasound imaging include a small field of view, speckled
image quality, attenuation-induced imaging artifacts and the need for intense operator skills.
Block-based attenuation compensation methods, widely employed in commercial scanners, produce
results with resolutions limited by the block size.
We have developed a novel Backscatter-Contour-Attenuation (BCA) joint estimation method for
reconstructing the backscatter and attenuation fields in ultrasound images, along with automatic
structure boundary segmentation. The method is based upon a bio-mathematical model
relating the observed signal to the underlying attributive backscatter and attenuation sources.
We solve the mathematical inverse problem by formulating a functional minimization with region-based isotropic regularizations.
|
|
|
Attenuation compensation of simulated images with uncorrelated speckle.
(a) Attenuated images with ellipses as initial contours,
(b) initial attenuation fields,
(c) attenuation compensated images with noise remained,
(d) reconstructed attenuation field.
|
|
Simulated images with fully developed, spatially correlated speckle.
(a) Simulated B-scans with ellipses as initial contours,
(b) attenuation compensated images with speckle retained,
(c) extracted backscatter fields,
(d) extracted attenuation fields.
|
|
|
FIELD II simulation.
(a) B-scan of a cylindrical inclusion with +10 dB backscatter and +1.0 dB/cm/MHz attenuation,
(b) conventional angular compounding,
(c) BCA attenuation compensated image with speckle remained,
(d) extracted backscatter field.
|
|
Olive phantom.
(a) a longitudinal B-scan of an olive embedded in jelly mixed with flour,
(b) conventional angular compounding,
(c) BCA attenuation compensated image,
(d) extracted backscatter field.
|
|
