A blockwise relaxation labeling scheme and its application to edge detection in cardiac MR image sequences
✍ Scribed by Tülay Adalı; Yue Wang; Nidhi Gupta
- Publisher
- John Wiley and Sons
- Year
- 1998
- Tongue
- English
- Weight
- 431 KB
- Volume
- 9
- Category
- Article
- ISSN
- 0899-9457
No coin nor oath required. For personal study only.
✦ Synopsis
We present a segmentation scheme for magnetic resoartifacts, these effects are due to the respiratory cycles of the nance (MR) image sequences based on vector quantization of a heart resulting in blurring of the edges of the ventricle. In dark block-partitioned image followed by a relaxation labeling procedure.
blood cine acquisitions [2], static blood might give out a very
By first searching a coarse segmentation, the algorithm yields very strong signal, whereas fresh, moving blood might not give any fast and effective performance on images that are inherently noisy, signal. Because of extended exposure, problems such as tissue and can effectively use the correlation in a sequence of images for saturation and absence of edge signal on segments of the myocarrobust performance and efficient implementation. The algorithm dedial wall are encountered. Besides the problems posed by the fines feature vectors by the local histogram on a block-partioned artifacts, in general, the images have a low signal-to-noise ratio image and approximates the local histograms by normal distributions.
(SNR), high speckle noise, low spatial resolution, and high pixel
The relative entropy is chosen as the meaningful distance measure between the feature vectors and the templates. After initial computa-intensity variability. The volume of data in a complete cardiac tion of the normal distribution parameters, a blockwise classification study can be immense. Processing these data manually for border maximization algorithm classifies blocks in the block-partitioned imidentification is a very time-consuming, tedious, and expensive age by minimizing their relative entropy distance for a coarse-resoluprocess, and presents problems of inter-and intraobserver varition segmentation; and finally, finer resolution is obtained by contexability. The development of algorithms that provide automatic tual Bayesian relaxation labeling in which label update is performed analysis of the acquired information would be very beneficial.
pixelwise by incorporating neighborhood information. Sequence pro-Among the approaches to the problem of endocardial contour cessing is then performed to segment all images in the sequence.
determination, Zhang and Geiser [3] proposed an algorithm for
The scheme is applied to left ventricular boundary detection in shortdetecting endocardial borders from echocardiograms where rough axis MR image sequences, and results are presented to show that estimates of the borders and radii of the ventricles are defined the algorithm successfully extracts the endocardial contours and that sequence processing significantly improves edge detection perfor-by an operator. Temporal co-occurrence matrices for regional mance and can avoid local minima problems.