Konstantinos K. Delibasis

Automatic retinal image registration scheme using global optimization techniques

Retinal image registration is commonly required in order to combine the complementary information in different retinal modalities. In this paper, a new automatic scheme to register retinal images is presented and is currently tested in a clinical environment. The scheme considers the suitability and efficiency of different image transformation models and function optimization techniques, following an initial preprocessing stage. Three different transformation models-affine, bilinear and projective-as well as three optimization techniques-downhill simplex method, simulated annealing and genetic algorithms-are investigated and compared in terms of accuracy and efficiency. The registration of 26 pairs of fluoroscein angiography and indocyanine green chorioangiography images with the corresponding red-free retinal images, showed the superiority of combining genetic algorithms with the affine and bilinear transformation models. A comparative study of the proposed automatic registration scheme against the manual method, commonly used in clinical practice, is finally presented showing the advantage of the proposed automatic scheme in terms of accuracy and consistency.

Multimodal registration of retinal images using self organizing maps

In this paper, an automatic method for registering multimodal retinal images is presented. The method consists of three steps: the vessel centerline detection and extraction of bifurcation points only in the reference image, the automatic correspondence of bifurcation points in the two images using a novel implementation of the self organizing maps and the extraction of the parameters of the affine transform using the previously obtained correspondences. The proposed registration algorithm was tested on 24 multimodal retinal pairs and the obtained results show an advantageous performance in terms of accuracy with respect to the manual registration.

Automatic model-based tracing algorithm for vessel segmentation and diameter estimation

An automatic algorithm capable of segmenting the whole vessel tree and calculate vessel diameter and orientation in a digital ophthalmologic image is presented in this work. The algorithm is based on a parametric model of a vessel that can assume arbitrarily complex shape and a simple measure of match that quantifies how well the vessel model matches a given angiographic image. An automatic vessel tracing algorithm is described that exploits the geometric model and actively seeks vessel bifurcation, without user intervention. The proposed algorithm uses the geometric vessel model to determine the vessel diameter at each detected central axis pixel. For this reason, the algorithm is fine tuned using a subset of ophthalmologic images of the publically available DRIVE database, by maximizing vessel segmentation accuracy. The proposed algorithm is then applied to the remaining ophthalmological images of the DRIVE database. The segmentation results of the proposed algorithm compare favorably in terms of accuracy with six other well established vessel detection techniques, outperforming three of them in the majority of the available ophthalmologic images. The proposed algorithm achieves subpixel root mean square central axis positioning error that outperforms the non-expert based vessel segmentation, whereas the accuracy of vessel diameter estimation is comparable to that of the non-expert based vessel segmentation.

A comparative study of surface‐ and volume‐based techniques for the automatic registration between CT and SPECT brain images

Image registration of multimodality images is an essential task in numerous applications in three-dimensional medical image processing. Medical diagnosis can benefit from the complementary information in different modality images. Surface-based registration techniques, while still widely used, were succeeded by volume-based registration algorithms that appear to be theoretically advantageous in terms of reliability and accuracy. Several applications of such algorithms for the registration of CT-MRI, CT-PET, MRI-PET, and SPECT-MRI images have emerged in the literature, using local optimization techniques for the matching of images. Our purpose in this work is the development of automatic techniques for the registration of real CT and SPECT images, based on either surface- or volume-based algorithms. Optimization is achieved using genetic algorithms that are known for their robustness. The two techniques are compared against a well-established method, the Iterative Closest Point-ICP. The correlation coefficient was employed as an independent measure of spatial match, to produce unbiased results. The repeated measures ANOVA indicates the significant impact of the choice of registration method on the magnitude of the correlation (F = 4.968, p = 0.0396). The volume-based method achieves an average correlation coefficient value of 0.454 with a standard deviation of 0.0395, as opposed to an average of 0.380 with a standard deviation of 0.0603 achieved by the surface-based method and an average of 0.396 with a standard deviation equal to 0.0353 achieved by ICP. The volume-based technique performs significantly better compared to both ICP (p<0.05, Neuman Keuls test) and the surface-based technique (p<0.05, Neuman-Keuls test). Surface-based registration and ICP do not differ significantly in performance.

Thoracic non-rigid registration combining self-organizing maps and radial basis functions

An automatic three-dimensional non-rigid registration scheme is proposed in this paper and applied to thoracic computed tomography (CT) data of patients with stage III non-small cell lung cancer (NSCLC). According to the registration scheme, initially anatomical set of points such as the vertebral spine, the ribs, and shoulder blades are automatically segmented slice by slice from the two CT scans of the same patient in order to serve as interpolant points. Based on these extracted features, a rigid-body transformation is then applied to provide a pre-registration of the data. To establish correspondence between the feature points, the novel application of the self-organizing maps (SOMs) is adopted. In particular, the automatic correspondence of the interpolant points is based on the initialization of the Kohonen neural network model capable to identify 500 corresponding pairs of points approximately in the two CT sets. Then, radial basis functions (RBFs) using the shifted log function is subsequently employed for elastic warping of the image volume, using the correspondence between the interpolant points, as obtained in the previous phase. Quantitative and qualitative results are also presented to validate the performance of the proposed elastic registration scheme resulting in an alignment error of 6 mm, on average, over 15 CT paired datasets. Finally, changes of the tumor volume in respect to each reference dataset are estimated for all patients, which indicate inspiration and/or movement of the patient during acquisition of the data. Thus, the practical implementation of this scheme could provide estimations of lung tumor volumes during radiotherapy treatment planning.

Detection of glaucomatous change based on vessel shape analysis

Glaucoma, a leading cause of blindness worldwide, is a progressive optic neuropathy with characteristic structural changes in the optic nerve head and concomitant visual field defects. Ocular hypertension (i.e. elevated intraocular pressure without glaucoma) is the most important risk factor to develop glaucoma. Even though a number of variables, including various optic disc and visual field parameters, have been used in order to identify early glaucomatous damage, there is a need for computer-based methods that can detect early glaucomatous progression so that treatment to prevent further progression can be initiated. This paper is focused on the description of a system based on image processing and classification techniques for the estimation of quantitative parameters to define vessel deformation and the classification of image data into two classes: patients with ocular hypertension who develop glaucomatous damage and patients with ocular hypertension who remain stable. The proposed system consists of the retinal image preprocessing module for vessel central axis segmentation, the automatic retinal image registration module based on a novel application of self organizing maps (SOMs) to define automatic point correspondence, the retinal vessel attributes calculation module to select the vessel shape attributes and the data classification module, using an artificial neural network classifier, to perform the necessary subject classification. Implementation of the system to optic disc data from 127 subjects obtained by a fundus camera at regular intervals provided a classification rate of 87.5%, underscoring the value of the proposed system to assist in the detection of early glaucomatous change.

Enhancing classification accuracy utilizing globules and dots features in digital dermoscopy

The interest in image dermoscopy has been significantly increased recently and skin lesion images are nowadays routinely acquired for a number of skin disorders. An important finding in the assessment of a skin lesion severity is the existence of dark dots and globules, which are hard to locate and count using existing image software tools. In this work we present a novel methodology for detecting/segmenting and count dark dots and globules from dermoscopy images. Segmentation is performed using a multi-resolution approach based on inverse non-linear diffusion. Subsequently, a number of features are extracted from the segmented dots/globules and their diagnostic value in automatic classification of dermoscopy images of skin lesions into melanoma and non-malignant nevus is evaluated. The proposed algorithm is applied to a number of images with skin lesions with known histo-pathology. Results show that the proposed algorithm is very effective in automatically segmenting dark dots and globules. Furthermore, it was found that the features extracted from the segmented dots/globules can enhance the performance of classification algorithms that discriminate between malignant and benign skin lesions, when they are combined with other region-based descriptors.

MR functional cardiac imaging: segmentation, measurement and WWW based visualisation of 4D data

Abstract This paper considers the problem of ventricular segmentation and visualisation from dynamic (4D) MR cardiac data covering an entire patient cardiac cycle, in a format that is compatible with the web. Four different methods are evaluated for the process of segmentation of the objects of interest: The K-means clustering algorithm, the fuzzy K-means (FKM) algorithm, self-organizing maps (SOMs) and seeded region growing algorithm. The technique of active surface is then subsequently applied to refine the segmentation results, employing a deformable generalised cylinder as geometric primitive. The final ventricular models are presented in VRML 2.0 format. The same process is repeated for all the 3D volumes of the cardiac cycle. The radial displacement between end systole and end diastole is calculated for each point of the active surface and is encoded in colour on the VRML vertex, using the RGB colour model. Using the VRML 2.0 specifications, morphing is performed showing all cardiac phases in real time. The expert has the ability to view the objects and interact with them using a simple internet browser. Preliminary results of normal and abnormal cases indicate that very important pathological situations (such as infarction) can be visualised and thus easily diagnosed and localised with the assistance of the proposed technique.

Refinement of human silhouette segmentation in omni-directional indoor videos

Abstract In this paper, we present a methodology for refining the segmentation of human silhouettes in indoor videos acquired by fisheye cameras. This methodology is based on a fisheye camera model that employs a spherical optical element and central projection. The parameters of the camera model are determined only once (during calibration), using the correspondence of a number of user-defined landmarks, both in real world coordinates and on a captured video frame. Subsequently, each pixel of the video frame is inversely mapped to the direction of view in the real world and the relevant data are stored in look-up tables for fast utilization in real-time video processing. The proposed fisheye camera model enables the inference of possible real world positions and conditionally the height and width of a segmented cluster of pixels in the video frame. In this work we utilize the proposed calibrated camera model to achieve a simple geometric reasoning that corrects gaps and mistakes of the human figure segmentation, detects segmented human silhouettes inside and outside the room and rejects segmentation that corresponds to non-human activity. Unique labels are assigned to each refined silhouette, according to their estimated real world position and appearance and the trajectory of each silhouette in real world coordinates is estimated. Experimental results are presented for a number of video sequences, in which the number of false positive pixels (regarding human silhouette segmentation) is substantially reduced as a result of the application of the proposed geometry-based segmentation refinement.

Automatic point correspondence using an artificial immune system optimization technique for medical image registration.

In this paper, an automatic method for determining pairs of corresponding points between medical images is proposed. The method is based on the implementation of an artificial immune system (AIS). AIS is a relatively novel, population based category of algorithms, inspired by theoretical immunologic models. When used as function optimizers, AIS have the attractive property of locating the global optimum of a function as well as a large number of strong local optimum points. In this work, AIS has been applied both for the extraction of an optimal set of candidate points on the reference image and the definition of their corresponding ones on the second image. The performance of the proposed AIS algorithm is evaluated against the widely used Iterative Closest Point (ICP) algorithm in terms of the accuracy of the obtained correspondences and in terms of the accuracy of the point-based registration by the two correspondence algorithms and the Mutual Information criterion, as an intensity-based registration method. Qualitative and quantitative results involving 92 X-ray dental and 10 retinal image pairs subject to known and unknown transformations are presented. The results indicate a superior performance of the proposed AIS algorithm with respect to the ICP algorithm and the Mutual Information, in terms of both correct correspondence and registration accuracy.