Radek Benes

Novel Method for Localization of Common Carotid Artery Transverse Section in Ultrasound Images Using Modified Viola-Jones Detector

This article describes a novel method for highly accurate and effective localization of the transverse section of the carotis comunis artery in ultrasound images. The method has a high success rate, approximately 97%. Unlike analytical methods based on geometric descriptions of the object sought, the method proposed here can cover a large area of shape variation of the artery under study, which normally occurs during examinations as a result of the pressure on the examined tissue, tilt of the probe, setup of the sonographic device, and other factors. This method shows great promise in automating the process of determining circulatory system parameters in the non-invasive clinical diagnostics of cardiovascular diseases. The method employs a Viola-Jones detector that has been specially adapted for efficient detection of transverse sections of the carotid artery. This algorithm is trained on a set of labeled images using the AdaBoost algorithm, Haar-like features and the Matthews coefficient. The training algorithm of the artery detector was modified using evolutionary algorithms. The method for training a cascade of classifiers achieves on a small number of positive and negative training data samples (about 500 images) a high success rate in a computational time that allows implementation of the detector in real time. Testing was performed on images of different patients for whom different ultrasonic instruments were used under different conditions (settings) so that the algorithm developed is applicable in general radiologic practice.

Circle detection in pulsative medical video sequence

The article deals with a new method for the detection of pulsative circular objects in a medical video sequence. The motivation for investigating this method consists in the fact that a circular object is not very apparent and its detection in such a frame is inaccurate. In some cases of medical images, the pulsative character of the circular area being searched can be used for its localisation. The proposed method starts from an analysis of movement, using optical flow estimation. The compensation of global movement is necessary because only local pulsative movement during the video sequence is assumed. The optical flow estimation is followed by another main processing step: the Hough Transform for the circle position estimation. Circles with expected properties are selected using the Bayes classifier. Finally, the circle position in a single frame is adapted using the analysis of average pixel intensity in the directions starting from the circle centre.

Multi-focus thermal image fusion

This paper proposes a novel algorithm for multi-focus thermal image fusion. The algorithm is based on local activity analysis and advanced pre-selection of images into fusion process. The algorithm improves the object temperature measurement error up to 5^oC. The proposed algorithm is evaluated by half total error rate, root mean squared error, cross correlation and visual inspection. To the best of our knowledge, this is the first work devoted to multi-focus thermal image fusion. For testing of proposed algorithm we acquire six thermal image set with objects at different focal depth.

Object localization in medical images

This paper is focused on localization of objects in medical images. A novel improvement of an existing method for localization of artery in longitudinal ultrasound B-mode scan is proposed in the paper. The localization is based on a classification of pixels according to image information in their neighborhood. To suppress misclassified points, a novel RANSAC based method is proposed. This method is able to find the most appropriate mathematical model of depicted common carotid artery (CCA) on the basis of previous classification. The proposed RANSAC based method with its mathematical footing is described in detail and the results of method within the algorithm for localization of artery are enclosed. By using the proposed RANSAC based method the localization method becomes very robust.

Automatic measurement of intima media thickness with preselection of the most suitable places

This paper describes a novel method for a robust segmentation of layers situated on the far wall of the common carotid artery. Thickness of innermost two layers, called intima-media thickness (IMT), is highly important marker predicting a risk of cardiovascular events. The novelty of the proposed methods resides in an automatic initial determination of suitable places along the artery, where the layers are the most visible and easy to segment. Then, the IMT is measured only in these places and not along whole artery as is performed in almost all state of the art methods. This enables to measure the IMT even in images where the layers are visible only in a part of image. This is one of the main benefits of the proposed method as well as its low computational burdens and low demands on precision of its initialization.

Robustness evaluation of corner detectors for use in ultrasound image processing

This article deals with evaluation of suitability of interest point detectors (specifically corner detectors) for utilization in ultrasound image processing. Namely, the Harris detector, FAST (Features from Accelerated Segment Test) detector and Kanade-Lucas-Tomasi (KLT) detector have been tested on two data sets of ultrasound (US) images. The evaluation process consists of three experiments in which the images have been artificially corrupted by different types of noise or their brightness has been uniformly changed. The FAST detector proved to be more robust against noise than Harris and KLT detectors whereas the Harris and KLT detector outperformed FAST detector in case of brightness change.

Towards an Optimal Interest Point Detector for Measurements in Ultrasound Images

Abstract This paper focuses on the comparison of different interest point detectors and their utilization for measurements in ultrasound (US) images. Certain medical examinations are based on speckle tracking which strongly relies on features that can be reliably tracked frame to frame. Only significant features (interest points) resistant to noise and brightness changes within US images are suitable for accurate long-lasting tracking. We compare three interest point detectors - Harris-Laplace, Difference of Gaussian (DoG) and Fast Hessian - and identify the most suitable one for use in US images on the basis of an objective criterion. Repeatability rate is assumed to be an objective quality measure for comparison. We have measured repeatability in images corrupted by different types of noise (speckle noise, Gaussian noise) and for changes in brightness. The Harris-Laplace detector outperformed its competitors and seems to be a sound option when choosing a suitable interest point detector for US images. However, it has to be noted that Fast Hessian and DoG detectors achieved better results in terms of processing speed.