University of Washington

Robert M Haralick

Florence H Sheehan

Paul D Sampson

Jasjit Suri

Robert E Johnson

Yong Shao

Arun K Somani

Kenneth H Cooper

Chung-Ho Chen

Chang-Kyu Lee

Craig M Wittenbrink

A method for defining a three-dimensional surface of at least a portion of a patient's heart, based on data obtained by ultrasound imaging of the heart. The imaging is carried out over at least one cardiac cycle and provides a plurality of images in different image planes made with an ultrasound transducer at known positions and orientations. At least three anatomical landmarks in these images are manually identified in each of the images. An ultrasound mesh model in which a three-dimensional abstract mesh defining an archetype shape for a population of other hearts is developed. This abstract mesh has associated with it a covariance that defines the extent of variation of control vertices in the mesh for the population of training data used to derive the mesh model. The mesh model is rigidly aligned with the images of the patient's heart. Predicted images in planes corresponding to those of the images for the patient's heart and derived from the mesh model are compared corresponding images of the patient's heart. Control vertices are included with the abstract mesh for reiteratively adjusting a shape of the mesh model to optimize the fit of the predicted images to the observed images of the patient's heart. This adjustment and comparison continues until an acceptable fit is obtained, yielding an output that defines the shape of the endocardium and epicardium of the left ventricle or other portion of the patient's heart, in three dimensions, for use in determining cardiac parameters.

Automated delineation of heart contours from images using

An automated method for evaluating image data taken over a sequence of image frames to determine a contour of a left ventricle (LV). The sequence of image frames are converted to digital data that identify a gray scale value for each pixel in each image frame. An initial LV region is estimated based upon training data parameters derived from processing manually drawn contours of the LV in the hearts of other individuals, corresponding to the sequence of image frames being analyzed. Prior to producing the initial estimate, the location of the aortic valve in the sequence of image frames for the patient is determined to serve as a reference. The data derived frown the contours of the LVs for the other hearts are rotated and translated relative to this reference to insure that the training data information corresponds to the position and orientation of the LV being analyzed. In addition, the effects of tissue mass comprising a diaphragm adjacent to the inferior portion of the LV are compensated by masking out the diaphragm. Uncertainty of the contour of the LV in this portion due to the presence of the diaphragm is thereby minimized. The resulting automatically determined contours of each image frame can be displayed to enable a physician to more readily diagnose physiological defects of the heart.

Method for determining the contour of an in vivo organ using multiple

A method for automatically evaluating image data taking over a sequence of image frames to determine a contour of a left ventricle (LV). The sequence of image frames are converted to digital data that identify a gray scale value for each pixel in each image frame. Using probability, an initial LV region is estimated based upon previously determined a priori parameters that are derived from processing manually drawn contours of the LV in other hearts, for the sequence of image frames being analyzed. The initial estimate of the LV region is refined using motion constraints, which are also derived from training data determined from the manually-drawn contours of other hearts. An LV region is estimated and its coordinates are extracted for use in applying global shape constraints that provide a more accurate LV boundary estimate for the region around an apex of the left ventricle. Finally, the estimated LV boundary estimate is tested for consistency, yielding an accepted estimate, or warning that the estimate is unreliable. The resulting automatically determined contours of each image frame can be displayed almost immediately after the image data are produced so that a physician can more readily diagnose physiological defects of the heart.

A plurality of program processors, shared memory, dual port memory, external cache memory and a control processor form a multiprocessor system. A shared memory bus links the program processors, shared memory, dual port memory and external cache memory. Program processor I/O occurs through a pair of serial I/O channels coupled to one port of the dual port memory. A write generate mode is implemented for updating cache by first allocating lines of shared memory as write before read areas. For such lines, cache tags are updated directly on cache misses without reading from memory. A hit is forced for such line, resulting in valid data at the updated part and invalid data at the remaining portion. Thus, part of the line is written to and the rest invalidated. The invalid portions are not read, unless preceded by a write operation. The mode reduces the number of bus cycles by making write misses more efficient.

Growing community of inventors

Seattle, WA, United States of America

Robert M Haralick