TISSUE CHARACTERIZATION BY SPECTRAL ANALYSIS OF ULTRASONIC IMAGES

A new method has been developed to distinguish differences in microscopic tissue structure from noninvasive ultrasound B-mode video images. The power spectrum of the processed ultrasonic speckle pattern provides information about the distribution of microreflectors in the tissue. Spectra are compared with the Difference Cross-Covariance to indicate differences in tissue structures.;Speckled ultrasound is detected, filtered and sampled to form a B-mode video image. The spatial spectrum of the image is noisy and has a low spectral resolution, complicating determination of the spectral components. A new method generates ultrasound beam line spectra from the video image and uses the Difference Cross-Covariance to indicate differences between their composite shapes. The resulting difference curve indicates the difference between tissue structures.;The method is tested in several ways. A computer model of speckle patterns confirms the existence of two power spectral components with simulations of microreflectors spaced randomly, regularly and combinations of both. Tissue mimicking phantoms with glass bead microreflectors indicate sensitivity to changes in scatterer size but not concentration for a random distribution of scatterers. Imaging directly from the surface of the heart in situ indicates that the technique can detect changes in the structure of myocardium as ischemia increases. Finally, the Difference Cross-Covariance is used to distinguish between healthy and diseased myocardia in vivo with p ;The Difference Cross-Covariance comparison of ultrasonic B-mode video image spatial spectra is a robust technique for determining differences in microscopic tissue structure. An added advantage is that it can be performed with standard image processing hardware on archived video taped images.;Noninvasive medical imaging reveals the anatomic makeup of biological tissues but does not describe their microscopic structure. Ultrasonic B-mode imaging has a speckled appearance caused by interference of portions of the ultrasound wave front backscattered by microreflectors, which are discontinuities in density and compressibility smaller than the wavelength of ultrasound, in tissue. Ultrasonic speckle contains information about the distribution of microreflectors. Speckle pattern spectra can be divided into two components: the overall shape of the power spectrum, which is related to the ratio of random to regularly spaced scatterers, and local peaking, which is related to the spacing of regular scatterers.