Derivation of optimal filters for the detection of coronary arteries

A new approach to contour detection in x-ray arteriograms: the wavecontour

OBJECTIVES

We sought to develop a novel approach (the Wavecontour) for the detection of contours in vascular x-ray images, designed to eliminate any systematic underestimation or overestimation for vessel sizes in the range of 0.5 to 15 mm and further minimize the influence of the user-defined start points and end points.

MATERIALS AND METHODS

This method is based on the Wavefront Propagation principle in a 2-stage approach. Two validation experiments were performed: a Plexiglas phantom study (tube sizes ranging from 0.51 to 9.9 mm) and an in vivo patient study (114 patients with various degrees of stenosis).

RESULTS

The phantom study demonstrated an accuracy of 0.007 mm and a precision of 0.072 mm. The patient study showed a high similarity between the detected and the expert-drawn contours: 93% for a threshold of 1.0 pixel and 81% for a threshold of 0.5 pixels. Furthermore, the contours are robust in complex lesions and are almost independent in the middle part of the segment from the user-defined start point and end point. A variation of only 0.6 pixels exists in the middle 60% of the contours.

CONCLUSIONS

Our new Wavecontour approach performs very well on phantom images as well as on clinical data over the whole range of 0.5 to 15 mm and results in more robust QCA/QVA analyses.

Adaptive edge localisation approach for quantitative coronary analysis

Lack of reliability, user dissatisfaction and errors in determining coronary vessel wall characteristics are challenging issues in quantitative coronary analysis (QCA). A new approach is proposed for QCA that tackles these issues. The proposed approach extracts the coronary vessel edges by applying dynamic programming techniques that use human-based decision criteria, adaptive edge detection and feature-based cost minimisation. This approach uses image gradient, image intensity, boundary continuity and adaptive thresholding to gain maximum quality assurance. The validation of this approach was conducted through modelled phantoms and real X-ray angiograms. The results show that the accuracies obtained were 0.0116mm and 0.06mm, respectively, and the precisions were 0.0263mm, and 0.04mm, respectively. The proposed approach is reliable, reproducible and user friendly and provides high precision compared with recently published results. Furthermore, the significance of the proposed approach and its limitations are also discussed.

Detection and measurement of retinal vessels in fundus images using amplitude modified second-order Gaussian filter

In this paper, the fitness of estimating vessel profiles with Gaussian function is evaluated and an amplitude-modified second-order Gaussian filter is proposed for the detection and measurement of vessels. Mathematical analysis is given and supported by a simulation and experiments to demonstrate that the vessel width can be measured in linear relationship with the "spreading factor" of the matched filter when the magnitude coefficient of the filter is suitably assigned. The absolute value of vessel diameter can be determined simply by using a precalibrated line, which is typically required since images are always system dependent. The experiment shows that the inclusion of the width measurement in the detection process can improve the performance of matched filter and result in a significant increase in success rate of detection.

A new model-based technique for enhanced small-vessel measurements in X-ray ciné-angiograms

Arterial diameter estimation from X-ray ciné angiograms is important for quantifying coronary artery disease (CAD) and for evaluating therapy. However, diameter measurement in vessel cross sections < or =1.0 mm is associated with large measurement errors. We present a novel diameter estimator which reduces both magnitude and variability of measurement error. We use a parametric nonlinear imaging model for X-ray ciné angiography and estimate unknown model parameters directly from the image data. Our technique allows us to exploit additional diameter information contained within the intensity profile amplitude, a feature which is overlooked by existing methods. This method uses a two-step procedure: the first step estimates the imaging model parameters directly from the angiographic frame and the second step uses these measurements to estimate the diameter of vessels in the same image. In Monte-Carlo simulation over a range of imaging conditions, our approach consistently produced lower estimation error and variability than conventional methods. With actual X-ray images, our estimator is also better than existing methods for the diameters examined (0.4-4.0 mm). These improvements are most significant in the range of narrow vessel widths associated with severe coronary artery disease.

Model-based morphological segmentation and labeling of coronary angiograms

A method for extraction and labeling of the coronary arterial tree (CAT) using minimal user supervision in single-view angiograms is proposed. The CAT structural description (skeleton and borders) is produced, along with quantitative information for the artery dimensions and assignment of coded labels, based on a given coronary artery model represented by a graph. The stages of the method are: 1) CAT tracking and detection; 2) artery skeleton and border estimation; 3) feature graph creation; and iv) artery labeling by graph matching. The approximate CAT centerline and borders are extracted by recursive tracking based on circular template analysis. The accurate skeleton and borders of each CAT segment are computed, based on morphological homotopy modification and watershed transform. The approximate centerline and borders are used for constructing the artery segment enclosing area (ASEA), where the defined skeleton and border curves are considered as markers. Using the marked ASEA, an artery gradient image is constructed where all the ASEA pixels (except the skeleton ones) are assigned the gradient magnitude of the original image. The artery gradient image markers are imposed as its unique regional minima by the homotopy modification method, the watershed transform is used for extracting the artery segment borders, and the feature graph is updated. Finally, given the created feature graph and the known model graph, a graph matching algorithm assigns the appropriate labels to the extracted CAT using weighted maximal cliques on the association graph corresponding to the two given graphs. Experimental results using clinical digitized coronary angiograms are presented.