Boris Cigale

SEGMENTATION OF OVARIAN ULTRASOUND IMAGES USING CELLULAR NEURAL NETWORKS

Segmentation of ovarian ultrasound images using cellular neural networks (CNNs) is studied in this paper. The segmentation method consists of five successive steps where the first four uses CNNs. In the first step, only rough position of follicles is determined. In the second step, the results are improved by expansion of detected follicles. In the third step, previously undetected inexpressive follicles are determined, while the fourth step detects the position of ovary. All results are joined in the fifth step. The templates for CNNs were obtained by applying genetic algorithm. The segmentation method has been tested on 50 ovarian ultrasound images. The recognition rate of follicles was around 60% and misidentification rate was around 30%.

Cost-efficient speckle interferometry with plastic optical fiber for unobtrusive monitoring of human vital signs

Abstract. A cost-efficient plastic optical fiber (POF) system for unobtrusive monitoring of human vital signs is presented. The system is based on speckle interferometry. A laser diode is butt-coupled to the POF whose exit face projects speckle patterns onto a linear optical sensor array. Sequences of acquired speckle images are transformed into one-dimensional signals by using the phase-shifting method. The signals are analyzed by band-pass filtering and a Morlet-wavelet-based multiresolutional approach for the detection of cardiac and respiratory activities, respectively. The system is tested with 10 healthy nonhospitalized persons, lying supine on a mattress with the embedded POF. Experimental results are assessed statistically: precisions of 98.8%±1.5% and 97.9%±2.3%, sensitivities of 99.4%±0.6% and 95.3%±3%, and mean delays between interferometric detections and corresponding referential signals of 116.6±55.5 and 1299.2±437.3  ms for the heartbeat and respiration are obtained, respectively.

Computerized detection and recognition of follicles in ovarian ultrasound images: a review

Observing changes in females’ ovaries is essential in obstetrics and gynaecological imaging, e.g., genetic engineering and human reproduction. It is particularly important to monitor the dynamics of ovarian follicles’ growth, as only fully mature and grown follicles, i.e., the dominant follicles have a potential to ovulate at the end of a follicular phase. Gynaecologists follow this process in two dimensions, but recently three-dimensional (3-D) ultrasound examinations are coming to the fore. This paper surveys the existing computer methods for detection, recognition, and analyses of follicles in two-dimensional (2-D) and 3-D ovarian ultrasound recordings. Our study focuses on the efficiency, validation, and assessment of proposed follicle processing algorithms. The most important processing steps were identified in order to compare their performances. Higher ranking solutions are suggested for the so-called best algorithm for 2-D and 3-D ultrasound recordings of ovarian follicles. Finally, some guidelines for future research in this field are discussed, in particular for 3-D ultrasound volumes.

Segmentation of Ovarian Ultrasound Images Using Single Template Cellular Neural Networks Trained with Support Vector Machines

Various applications of cellular neural networks (CNNs) on complex image processing tasks raise questions about an appropriate selection of template elements that determine the CNNs behavior. In previous research we utilized multiple time variant template elements cellular neural networks for segmentation, which has many advantages compared to conventional search space approaches. In this paper a novel approach which utilizes the formalism of support vector machines (SVMs) that utilizes only single time invariant set of CNN template elements, is introduced. The main advantage of this approach is reduction of the number of CNNs templates and confirms to the conventional applications of CNN.

Segmentation of ovarian ultrasound images using cellular neural networks trained by support vector machines

Various applications of cellular neural networks (CNNs) on complex image processing tasks raise questions about an appropriate selection of template elements that determine the CNNs behaviour. There are two possibilities: either to resort to the existing and published templates suitable for the problem under consideration or to construct the templates by one of well-known training methods, such as genetic algorithms, simulated annealing, etc. In this paper, a novel approach which utilizes the formalism of support vector machines (SVMs) is introduced. We found the CNN template optimisation done by this machine learning technique superior to other training methods. The learning time reduced from several hours to less than a minute. Testing our novel approach on ultrasound ovarian images, the obtained segmentation results and recognition rates for ovarian follicles were significantly better than with comparable solutions.

Obtaining completely stable cellular neural network templates for ultrasound image segmentation

Cellular neural networks (CNNs) have been successfully applied to image segmentation problem. Nevertheless, the main difficulty remains in the process of creating appropriate templates to solve a segmentation problem. In this paper we present machine learning approach to obtain completely stable CNN templates and compare the obtained results to unconstrained machine learning approach. Despite introduced constraints of templates stability the results are comparable to unobstructed ones.

Application of simulated annealing to biosignal classification and segmentation

Simulated annealing is no doubt one of the most popular optimisation methods. Like similar methods it is based on natural phenomena. In the paper two examples of how to use this method in medical applications are presented. In the first example the successfulness of simulated annealing is compared to a genetic algorithm on the problem of the optimal coefficients determination for cellular neural networks, when used for segmentation of ovarian ultrasound images. In this field the latter method produced better results. The second example discusses the usage of simulated annealing for optimisation of signal classification.

Short-Term Photoplethysmography Embedded in Household Appliances

This paper opens new insights into short-term photoplethysmography (PPG) in dynamic condition when opening a refrigerator. Five hundred twelve PPG signals from fingers of 7 healthy volunteers were acquired simultaneously and analysed for heartbeats. Parallel referential electrocardiogram recordings confirmed the heartbeat detection sensitivity and precision of 97.62% and 90.31%, respectively. We also assessed the mean blood pressure, firstly, by measuring pulse transit times and, secondly, by our novel method based on the blood inflow times at the door opening instants. The models derived with exponential fitting proved the inflow times a robust and accurate blood pressure indicator; the mean absolute error was 1.2±0.9 mmHg.

Fast Segmentation of Ovarian Ultrasound Volumes Using Support Vector Machines and Sparse Learning Sets

Ovarian ultrasound imaging has recently drawn attention because of the improved ultrasound-based diagnostic methods and because of its application to in-vitro fertilisation and prediction of women’s fertility. Modern ultrasound devices enable frequent examinations and sophisticated built-in image processing options. However, precise detection of different ovarian structures, in particular follicles and their growth still need additional, mainly off-line processing with highly specialised algorithms. Manual annotation of a whole 3D ultrasound volume consisting of 100 and more slices, i.e. 2D ultrasound images, is a tedious task even when using handy, computer-assisted segmentation tools. Our paper reveals how an application of support vector machines (SVM) can ease the follicle detection by speeding up the learning and annotation processes at the same time. An iterative SVM approach is introduced using training on sparse learning sets only. The recognised follicles are compared to the referential expert readings and to the results obtained after learning on the entire annotated 3D ovarian volume.

Directional 3D Wavelet Transform Based on Gaussian Mixtures for the Analysis of 3D Ultrasound Ovarian Volumes

The success of in-vitro fertilization can be predicted by a correct quantitative and qualitative assessment of ovarian follicles. Several ovarian follicle detection and recognition algorithms have been published. Their effectiveness is inferior to human follicle annotations due to various kinds of noise, degradations, and artefacts in ultrasonic images. This paper deals with an approach to recognize antral follicles from 2 mm in diameter in 3D ultrasound data. Its detection phase looks for candidate follicular regions, while the recognition phase assesses the likelihood of a region to correspond to a follicle. Three innovative definitions underpin the detection: Laplacian-of-Gaussian-based directional 3D wavelet transform, adaptive multiscale search based on Gaussian mixtures, and recursive convexity-based region splitting. A likelihood index is also introduced to support follicle recognition. The proposed approach was tested on 30 ultrasound ovarian volumes generated by different sonographic machines in stimulated and non-stimulated examination cycles. The obtained follicle recognition rates exceed those of the best 3D approaches known by about 10 percent, while qualitative assessments yield comparable values.