Jorge Alves Silva

Segmentation of the carotid intima-media region in B-mode ultrasound images

This paper proposes a new approach for the segmentation of both near-end and far-end intima-media regions of the common carotid artery in ultrasound images. The method requires minimal user interaction and is able to segment the near-end wall in arteries with large, hypoechogenic and irregular plaques, issues usually not considered previously due to the increased segmentation difficulty. The adventitia is detected by searching for the best fit of a cubic spline to edges having features compatible with the adventitia boundary. The algorithm uses a global smoothness constraint and integrates discriminating features of the adventitia to reduce the attraction by other edges. Afterwards, using the information of the adventitia location, the lumen boundary is detected by combining dynamic programming, smooth intensity thresholding surfaces and geometric snakes. Smooth contours that correctly adapt to the intima are produced, even in the presence of deep concavities. Moreover, unlike balloon-based snakes, the propagation force does not depend on gradients and does not require a predefined direction. An extensive statistical evaluation is computed, using a set of 47 images from 24 different symptomatic patients, including several classes, sizes and shapes of plaques. Bland-Altman plots of the mean intima-media thickness, for manual segmentations of two medical experts, show a high intra-observer and inter-observer agreement, with mean differences close to zero (mean between -0.10mm and 0.18mm) and with the large majority of differences within the limits of agreement (standard deviation between 0.10mm and 0.12mm). Similar plots reveal a good agreement between the automatic and the manual segmentations (mean between -0.07mm and 0.11mm and standard deviation between 0.11mm and 0.12mm).

Segmentation of ultrasound images of the carotid using RANSAC and cubic splines

A new algorithm is proposed for the semi-automatic segmentation of the near-end and the far-end adventitia boundary of the common carotid artery in ultrasound images. It uses the random sample consensus method to estimate the most significant cubic splines fitting the edge map of a longitudinal section. The consensus of the geometric model (a spline) is evaluated through a new gain function, which integrates the responses to different discriminating features of the carotid boundary: the proximity of the geometric model to any edge or to valley shaped edges; the consistency between the orientation of the normal to the geometric model and the intensity gradient; and the distance to a rough estimate of the lumen boundary. A set of 50 longitudinal B-mode images of the common carotid and their manual segmentations performed by two medical experts were used to assess the performance of the method. The image set was taken from 25 different subjects, most of them having plaques of different classes (class II to class IV), sizes and shapes. The quantitative evaluation showed promising results, having detection errors similar to the ones observed in manual segmentations for 95% of the far-end boundaries and 73% of the near-end boundaries.

Automatic detection of the carotid lumen axis in B-mode ultrasound images.

A new approach is introduced for the automatic detection of the lumen axis of the common carotid artery in B-mode ultrasound images. The image is smoothed using a Gaussian filter and then a dynamic programming scheme extracts the dominant paths of local minima of the intensity and the dominant paths of local maxima of the gradient magnitude with the gradient pointing downwards. Since these paths are possible estimates of the lumen axis and the far wall of a blood vessel, respectively, they are grouped together into pairs. Then, a pattern of two features is computed from each pair of paths and used as input to a linear discriminant classifier in order to select the pair of paths that correspond to the common carotid artery. The estimated lumen axis is the path of local minima of the intensity that belongs to the selected pair of paths. The proposed method is suited to real time processing, no user interaction is required and the number of parameters is minimal and easy to determine. The validation was performed using two datasets, with a total of 199 images, and has shown a success rate of 99.5% (100% if only the carotid regions for which a ground truth is available are considered). The datasets have a large diversity of images, including cases of arteries with plaque and images with heavy noise, text or other graphical markings inside the artery region.

Rehab@home: a tool for home-based motor function rehabilitation

Abstract Purpose: This paper presents the Rehab@home system, a tool specifically developed for helping neurological patients performing rehabilitation exercises at home, without the presence of a physiotherapist. It is centred on the rehabilitation of balance and on the sit-to-stand (STS) movement. Method: Rehab@home is composed of two Wii balance boards, a webcam and a computer, and it has two main software applications: one for patients to perform rehabilitation exercises and another one for therapists to visualize the data of the exercises. During the exercises, data from the boards and the webcam are processed in order to automatically assess the correctness of movements. Results: Rehab@home provides exercises for the rehabilitation of balance (in sitting and in standing positions), and for the execution of the STS movement. It gives automatic feedback to the patient and data are saved for future analysis. The therapist is able to adapt the difficulty of the exercises to match with each patients needs. A preliminary study with seven patients was conducted for evaluating their feedback. They appreciated using the system and felt the exercises more engaging than conventional therapy. Conclusions: Feedback from patients gives the hope that Rehab@home can become a great tool for complementing their rehabilitation process. Implications for Rehabilitation Rehab@home can be used at home by patients with motor deficits, without the presence of a therapist, as a complement to conventional therapy for accelerating the rehabilitation process. The system provides exercises for improving the balance and the STS movement capabilities of patients, gives automatic feedback, and saves video and load information from the movements for future analysis by the therapist. Its most important feature is adaptability: the therapist is able to tune the difficulty of the exercises for adapting them to the needs of each patient. Patients get more engaged for this type of exercises and think they can take profit from using it.

Segmentation of ultrasonic images of the carotid

A new algorithm for an effective and automatic segmentation of the carotid wall in ultrasonic images is proposed. It combines the speed of thresholding algorithms with the accuracy, flexibility and robustness of a successful geometric active contour model which incorporates an optimal image segmentation model in a level set framework. Due to the multiphase nature of these images, a sequential minimum cross entropy thresholding is used to get a first approximation of the segments, reducing the problem to a two phase segmentation. This thresholding solution is then used as a starting point for a two phase piecewise constant version of a geometric active contour model to reduce noise, smooth contours, improve their position accuracy and close eventual gaps in the carotid wall.

Improving Face Recognition by Video Spatial Morphing

The focus of this chapter is in the problem of using technology to grant access to restricted areas by authorised persons, hereafter called ‘clients’, and to deny access to unauthorised or unidentified persons, the so called ‘impostors’. Conventional methods, such as magnetic or smart cards, user/password login and others, are being progressively recognised as insecure due to their many shortcomings, like the possibility of being lost, damaged or forged. Other methods, particularly those based on biometrics, are being increasingly used as they allow the verification of an individual’s identity on the basis of precise and careful measures of biological and physiological characteristics, such as fingerprints, hand and palm print geometry, iris and retina patterns, voice and face recognition. Automatic face recognition has very much progressed in the last few years, making its use practical in experimental or commercial systems. However, further research is still needed to make these systems more robust, reliable and less dependant on special constraints, particularly those imposed on the data acquisition process. In order to be as flexible as possible, current face recognition systems must use a large database of facial views for each client, so that distinct poses and emotional states can be accommodated, as well as other short-term variations in appearance caused by cosmetics or beard size, and by the use of various accessories such as spectacles or earrings. These multiple views are intended to increase the individual’s recognition rate for the capture of a single facial test image. The large dimension of the faces database induces a number of problems, namely the requirement for more storage, the increased computing time for recognition and decision, and the need for more complex classifiers. In an attempt to cope with the above problems we have devised an alternative approach, essentially consisting in keeping a much smaller facial image database, and in testing for valid matches a number of images extracted from a video fragment acquired during the person’s path in direction to the protected entrance. The main advantages to be expected from this approach can be summarised as: (a) the size of the reference face database is substantially reduced, as a single image or a small number of images for each individual are kept, (b) the clients are not subject to much discomfort when building the database, as a single neutral view (for each relevant appearance) is

Calibration of a 3D data acquistion system using the ratio of two intensity images

The calibration steps of a 3-D data acquisition system based on the ratio of two intensity images are described. This is a triangulation-based structural light technique in which a multiple of projected light planes are identified by the ratio of two measured intensities. The calibration scheme can be divided in two main steps, which are described in detail: camera calibration and light plane calibration. Camera calibration consists of the determination of some geometric camera parameters that allow the determination of the line of sight of each pixel. Light plane calibration consists of the determination of the correspondence between the measured ratios and the position of the light planes. Some preliminary operations which ease the calculation of depth are also described. System accuracy is evaluated and results on test scenes are presented.

Automatic Characterization of the Serous Retinal Detachment Associated with the Subretinal Fluid Presence in Optical Coherence Tomography Images

Abstract An accurate detection of the macular edema (ME) presence constitutes a crucial ophthalmological issue as it provides useful information for the identification, diagnosis and treatment of different relevant ocular and systemic diseases. Serous Retinal Detachment (SRD) is a particular type of ME, which is characterized by the leakage of fluid that has a propensity of being accumulated in the macular region. This paper proposes a new methodology for the automatic identification and characterization of the SRD edema using Optical Coherence Tomography (OCT) images. The subretinal fluids and the External Limiting Membrane (ELM) retinal layers are identified and characterized to measure the disease severity. Four different visualization modules were designed including representative derived parameters to facilitate the doctor’s work in the diagnostic evaluation of ME. The different steps of this method were validated using the manual labelling provided by an expert clinician. The validation of the proposed method offered satisfactory results, constituting a suitable scenario with intuitive visual representations that also include different relevant biomarkers.

Parallel Asynchronous Strategies for the Execution of Feature Selection Algorithms

Reducing the dimensionality of datasets is a fundamental step in the task of building a classification model. Feature selection is the process of selecting a smaller subset of features from the original one in order to enhance the performance of the classification model. The problem is known to be NP-hard, and despite the existence of several algorithms there is not one that outperforms the others in all scenarios. Due to the complexity of the problem usually feature selection algorithms have to compromise the quality of their solutions in order to execute in a practicable amount of time. Parallel computing techniques emerge as a potential solution to tackle this problem. There are several approaches that already execute feature selection in parallel resorting to synchronous models. These are preferred due to their simplicity and capability to use with any feature selection algorithm. However, synchronous models implement pausing points during the execution flow, which decrease the parallel performance. In this paper, we discuss the challenges of executing feature selection algorithms in parallel using asynchronous models, and present a feature selection algorithm that favours these models. Furthermore, we present two strategies for an asynchronous parallel execution not only of our algorithm but of any other feature selection approach. The first strategy solves the problem using the distributed memory paradigm, while the second exploits the use of shared memory. We evaluate the parallel performance of our strategies using up to 32 cores. The results show near linear speedups for both strategies, with the shared memory strategy outperforming the distributed one. Additionally, we provide an example of adapting our strategies to execute the Sequential forward Search asynchronously. We further test this version versus a synchronous one. Our results revealed that, by using an asynchronous strategy, we are able to save an average of 7.5% of the execution time.

3D Mapping of Choroidal Thickness from OCT B-Scans

The choroid is the middle layer of the eye globe located between the retina and the sclera. It is proven that choroidal thickness is a sign of multiple eye diseases. Optical Coherence Tomography (OCT) is an imaging technique that allows the visualization of tomographic images of near surface tissues like those in the eye globe. The automatic calculation of the choroidal thickness reduces the subjectivity of manual image analysis as well as the time of large scale measurements.