Jan J. Gerbrands

Assessment of short-, medium-, and long-term variations in arterial dimensions from computer-assisted quantitation of coronary cineangiograms.

A computer-assisted technique has been developed to assess absolute coronary arterial dimensions from 35 mm cineangiograms. The boundaries of optically magnified and video-digitized coronary segments and the intracardiac catheter are defined by automated edge-detection techniques. Contour positions are corrected for pincushion distortion. The accuracy and precision of the edge detection procedure as assessed from cinefilms of contrast-filled acrylate (Perspex) models were -30 and 90 micrometers, respectively. The variability of the analysis procedure itself in terms of absolute arterial dimensions was less than 0.12 mm, and in terms of percentage arterial narrowing for coronary obstructions less than 2.74%. Short-, medium-, and long-term variability measurements were assessed from repeated coronary angiographic examinations performed 5 min, 1 hr, and 90 days apart, respectively. For all studies the mean differences in absolute diameters were less than 0.13 mm. The variability in obstruction diameter ranged from 0.22 mm for the best-controlled study (medium-term) to 0.36 mm for the least-controlled study (long-term); variability in reference diameter ranged from 0.15 to 0.66 mm, respectively. It is concluded that the biological variations are a source of major concern and that further attempts toward standardization of the angiographic procedure are seriously needed.

Coronary Artery Dimensions from Cineangiograms-Methodology and Validation of a Computer-Assisted Analysis Procedure

To evaluate the efficacy of modern therapeutic procedures in the catheterization laboratory, the effects of vasoactive drugs, as well as the effects of short and long term interventions on the regression or progression of coronary artery disease, an objective and reproducible technique for the assessment of coronary artery dimensions was developed. This paper describes the methodology of such a computer-assisted analysis system, as well as the results from a validation study on the accuracy and precision. A region in a 35 mm cineframe encompassing a selected arterial segment is optically magnified and converted into video format by means of a specially constructed cinevideo converter and digitized for subsequent analysis by computer. Contours of the arterial segment are detected automatically on the basis of first and second derivative functions. Contour data are corrected for pincushion distortion; arterial dimensions are presented in mm, where the calibration factor is derived from a computer-processed segment of the contrast catheter. The accuracy and precision of the edge detection procedure as assessed from cinefilms of perspex models (%-D stenosis ⩽70 percent) filled with contrast agent were -30 and 90 μm, respectively. The variablity of the analysis procedure by itself in terms of absolute arterial dimensions was less than 0.12 mm, and in terms of percentage arterial narrowing for coronary obstructions less than 2.74 percent. It is concluded that this system allows the measurement of coronary arterial dimensions in an objective and highly reproducible way.

ON THE RELATIONSHIPS BETWEEN SVD, KLT AND PCA

Abstract In recent literature on digital image processing much attention is devoted to the singular value decomposition (SVD) of a matrix. Many authors refer to the Karhunen-Loeve transform (KLT) and principal components analysis (PCA) while treating the SVD. In this paper we give definitions of the three transforms and investigate their relationships. It is shown that in the context of multivariate statistical analysis and statistical pattern recognition the three transforms are very similar if a specific estimate of the column covariance matrix is used. In the context of two-dimensional image processing this similarity still holds if one single matrix is considered. In that approach the use of the names KLT and PCA is rather inappropriate and confusing. If the matrix is considered to be a realization of a two-dimensional random process, the SVD and the two statistically defined transforms differ substantially.

Image sharpening by morphological filtering

Abstract This paper introduces a class of iterative morphological image operators with applications to sharpen digitized gray-scale images. It is proved that all image operators using a concave structuring function have sharpening properties. By using a Laplacian property, we introduce the underlying partial differential equation that governs this class of iterative image operators. The parameters of the operator can be determined on the basis of an estimation of the amount of blur present in the image. For discrete implementations of the operator class it is shown that operators using a parabolic structuring function have an efficient implementation and isotropic sharpening behavior.

Three-dimensional image segmentation using a split, merge and group approach

A 3-D segmentation algorithm is presented, based on a split, merge and group approach. It uses a mixed (oct/quad)tree implementation. A number of homogeneity criteria is discussed and evaluated. An example shows the segmentation of mythramycin stained cell nuclei.

Determination of optimal angiographic viewing angles: basic principles and evaluation study

Foreshortening of vessel segments in angiographic (biplane) projection images may cause misinterpretation of the extent and degree of coronary artery disease. The views in which the object of interest are visualized with minimum foreshortening are called optimal views. The authors present a complete approach to obtain such views with computer-assisted techniques. The object of interest is first visualized in two arbitrary views. Two landmarks of the object are manually defined in the two projection images. With complete information of the projection geometry, the vector representation of the object in the three-dimensional space is computed. This vector is perpendicular to a plane in which the views are called optimal. The user has one degree of freedom to define a set of optimal biplane views. The angle between the central beams of the imaging systems can be chosen freely. The computation of the orientation of the object and of corresponding optimal biplane views have been evaluated with a simple hardware phantom. The mean and the standard deviation of the overall errors in the calculation of the optimal angulation angles were 1.8 degrees and 1.3 degrees , respectively, when the user defined a rotation angle.

A network flow approach to reconstruction of the left ventricle from two projections

Abstract A new method for binary reconstruction of the left ventricle from two orthogonal projections is presented. A priori knowledge has to be incorporated to reduce the ambiguity of the problem. A minimum cost capacitated network flow algorithm is discussed, which yields the optimal solution with respect to the priori information. It is shown that this method can also be used in the presence of observation noise. The method is demonstrated by reconstructing several cross sections of a dogs left ventricle.

Locating facial features in image sequences using neural networks

The paper describes a method for the automatic location of facial features, such as eyes, nose and mouth, in image sequences using a neural network approach. It is shown that by modeling the feature sought as a structural assembly of micro-features, and by using a probabilistic interpretation of neural network outputs, it is possible to construct a location system that is more robust than a location system which uses the feature as a single entity. With this micro-feature approach, not only can the position of the features be found but shape of the features can be obtained as well.

Objective and quantitative segmentation evaluation and comparison

Abstract A general framework for segmentation evaluation is introduced after a brief review of previous work. The accuracy of object feature measurement is proposed as a criterion for judging the quality of segmentation results and assessing the performance of applied algorithms. This goal-oriented approach has been shown useful for an objective and quantitative study of segmentation techniques.

Transition region determination based thresholding

Abstract We present a newly developed thresholding technique which is not based on the images gray-level histogram. This technique is fully automatic and quite robust in the presence of noise and unexpected structures. Moreover, no empirical parameters are used, and no limitations on shape and size of objects are imposed. A comparison with histogram based threshold selection is also discussed.