Alexandru Paul Condurache

Fast Detection and Processing of Arbitrary Contrast Agent Injections in Coronary Angiography and Fluoroscopy

Percutaneous transluminal coronary angioplasty (PTCA) requires both pre-interventional cine-angiograms showing the contrasted vessel tree over several heart cycles, and live X-ray monitoring (fluoroscopy) during the catheterization. Navigation during the intervention can be facilitated by fusing the automatically synchronized cineangiogram with the interventional images, e.g. by overlaying the synchronized angiogram over the interventional images. Clearly, this fusion should be limited to those frames of the angiogram which show the full contrasted vessel tree. Conversely, if contrast agent appears in the fluoroscopy images, overlay is not required and should be switched off. To these ends, we describe approaches for the detection and processing of contrast agent injections in cardiac X-ray image sequences.

Fast and robust diaphragm detection and tracking in cardiac x-ray projection images

A number of image analysis tasks of the heart region have to cope with both the problem of respiration and heart contraction. While the heart contraction status can be estimated based on the ECG, respiration status estimation must be based on the images themselves, unless additional devices for respiration measurements are used. Since diaphragm motion is closely linked to respiration, we describe a method to detect and track the diaphragm in x-ray projections. We model the diaphragm boundary as being approximately circular. Diaphragm detection is then based on edge detection followed by a Hough transform for circles. To avoid that the detection algorithm is misled by high frequency image content, we first apply a morphological multi-scale top hat operator. A Canny edge detector is then applied to the top hat filtered images. In the edge images, the circle corresponding to the diaphragm boundary is found by the Hough transform. To restrict the search in the 3D Hough parameter space (parameters are circle center coordinates and radius), prior anatomical knowledge about position and size of the diaphragm for the given image acquisition geometry is taken into account. In subsequent frames, diaphragm position and size are predicted from previous detection and tracking results. For each detection result, a confidence measure is computed by analyzing the Hough parameter space with respect to the goodness of the peak giving the circle parameters and by analyzing the coefficient of variation of the pixel that form the circle described by the maximum in Hough parameter space. If the confidence is not sufficiently high -- indicating a poor fit between the Hough circle and true diaphragm boundary -- the detection result is optionally refined by an active contour algorithm.

Statistical-model-based identification of complete vessel-tree frames in coronary angiograms

Coronary angiograms are pre-interventionally recorded moving X-ray images of a patients beating heart, where the coronary arteries are made visible by a contrast medium. They serve to diagnose, e.g., stenoses, and as roadmaps during the intervention itself. Covering about three to four heart cycles, coronary angiograms consist of three underlying states: inflow, when the contrast medium flows into the vessels, filled state, when the whole vessel tree is visible and outflow, when the contrast medium is washed out. Obviously, only that part of the sequence showing the full vessel tree is useful as a roadmap. We therefore describe methods for automatic identification of these frames. To this end, a vessel map with enhanced vessels and compressed background is first computed. Vessel enhancement is based on the observation that vessels are the locally darkest oriented structures with significant motion. The vessel maps can be regarded as containing two classes, viz. (bright) vessels and (dark)background. From a histogram analysis of each vessel map image, a time-dependent feature curve is computed in which the states inflow, filled state and outflow can already visually be distinguished. We then describe two approaches to segment the feature curve into these states: the first method models the observations in each state by a polynomial, and seeks the segmentation which allows the best fit of three polynomials as measured by a Maximum-Likelihood criterion. The second method models the state sequence by a Hidden Markov model, and estimates it using the Maximum a Posteriori (MAP)-criterion. We will present results for a number of angiograms recorded in clinical routine.

Segmentation of retinal vessels with a hysteresis binary-classification paradigm.

Vessel segmentation in photographies of the retina is needed in a set of computer-supported medical applications related to diagnosis and surgery planning. Considering each pixel in an image as a point in a feature space, segmentation is a binary classification problem where pixels need to be assigned to one of two classes: object and background. We describe a paradigm of hysteresis-classifier design that we apply to the problem of vessel segmentation. Before classification, a multidimensional feature vector is computed for each pixel, such that in the corresponding feature space, vessels and background are more separable than in the original image space. Several classifiers that stem from the hysteresis-classifier design paradigm are tested with this feature space on publicly available databases. These classifiers are very fast and achieve results that are comparable or even superior to known dedicated methods. Hysteresis-based classifiers represent a fast and accurate solution for the retinal-vessel segmentation problem.

Vessel segmentation and analysis in laboratory skin transplant micro-angiograms

The success of skin transplantations depends on the adequate revascularization of the transplanted dermal matrix. To induce vessel growth or angiogenesis, pharmacological substances may be applied to the dermal matrix. The effectiveness of different such substances has been evaluated in laboratory experiments. For this purpose, the surface and length of newly grown vessels have to be measured in micro-angiograms (x-ray images of the blood vessels recorded after the injection of a radiopaque substance) of tissue transplanted on the back of laboratory animals. To this end we describe in this contribution a vessel analysis environment central to which is a vessel segmentation tool for surface quantification in fasciocutaneous skin transplant micro-angiograms.

A full skin defect model to evaluate vascularization of biomaterials in vivo.

Insufficient vascularization is considered to be one of the main factors limiting the clinical success of tissue-engineered constructs. In order to evaluate new strategies that aim at improving vascularization, reliable methods are required to make the in-growth of new blood vessels into bio-artificial scaffolds visible and quantify the results. Over the past couple of years, our group has introduced a full skin defect model that enables the direct visualization of blood vessels by transillumination and provides the possibility of quantification through digital segmentation. In this model, one surgically creates full skin defects in the back of mice and replaces them with the material tested. Molecules or cells of interest can also be incorporated in such materials to study their potential effect. After an observation time of ones own choice, materials are explanted for evaluation. Bilateral wounds provide the possibility of making internal comparisons that minimize artifacts among individuals as well as of decreasing the number of animals needed for the study. In comparison to other approaches, our method offers a simple, reliable and cost effective analysis. We have implemented this model as a routine tool to perform high-resolution screening when testing vascularization of different biomaterials and bio-activation approaches.

Vessel Segmentation in 2D-Projection Images Using a Supervised Linear Hysteresis Classifier

2D projection imaging is a widely used procedure for vessel visualization. For the subsequent analysis of the vasculature, precise measurements of e.g. vessel area, vessel length or the number of vessel segments are needed. To achieve these goals vessel enhancement and segmentation are required. While there are already many vasculature specific vessel segmentation algorithms, we describe in this contribution a more general supervised segmentation method which includes a feature extraction step followed by feature selection and segmentation based on the hysteresis classification paradigm. The method was tested on retina photographs. The rates of false positives and correct classifications were comparable with dedicated methods on similar data sets while it needed less time for both training and providing a segmentation result

A two-stage-classifier for defect classification in optical media inspection

In this paper we address the problem of inspecting optical media like compact disks and digital versatile disks. Here, defective disks have to be identified during production. For optimizing the production process and in order to be able to decide how critical a certain defect is, the defects found have to be classified. As this has to be done online, the classification algorithm has to work very fast. With regard to speed, the well known minimum distance classifier is usually a good choice. However, when training data are not well clustered in the feature-space this classifier becomes rather unreliable. To trade-off speed and reliability we propose a two-stage-algorithm. It combines the fast minimum distance classification with a reliable fuzzy k-nearest neighbor classifier. The resulting two-stage-classifier is considerably faster than the fuzzy k-nearest neighbor classifier. Its classification rates are in the range of the fuzzy k-nearest neighbor classifier and far better than those of the minimum distance classifier. To evaluate the results, we compare them to the results obtained using various standard classifiers.

Vessel Segmentation for Angiographic Enhancement and Analysis

Angiography is a widely used method of vessel imaging for the diagnosis and treatment of pathological manifestations as well as for medical research. Vessel segmentation in angiograms is useful for analysis but also as a means to enhance the vessels. Often the vessel surface has to be quantified to evaluate the success of certain drugs treatment (e.g. aimed at angiogenesis in the case of transplanted skin) or to gain insight into different pathological manifestations (e.g. proliferative diabetic retinopathy). In this paper we describe algorithms for automatic vessel segmentation in angiograms. We first enhance likely vessel regions to obtain a vessel map which is then segmented. To remove false positives we accept in a second step only those regions showing branchings and bifurcations which are typical for a vessel tree.

Automatic measuring of quality criteria for heart valves

Patients suffering from a heart valve deficiency are often treated by replacing the valve with an artificial or biological implant. In case of biological implants, the use of porcine heart valves is common. Quality assessment and inspection methods are mandatory to supply the patients (and also medical research) with only the best such xenograft implants thus reducing the number of follow-up surgeries to replace worn-up valves. We describe an approach for automatic in-vitro evaluation of prosthetic heart valves in an artificial circulation system. We show how to measure the orifice area during a heart cycle to obtain an orifice curve. Different quality parameters are then estimated on such curves.