Darius Jegelevičius

Model based investigation of retinal vessel tortuosity as a function of blood pressure: preliminary results

Tortuosity is one of parameters which describe a state of the eye fundus blood vessels. Tortuosity can be estimated from the detected vessels in optical fundus images. The increase in vessel tortuosity was observed in eyes of patients with advanced background diabetic retinopathy, papilloedema, even in some completely healthy eyes (in this case tortuosity does not change in time). Though many methods to estimate eye vessel tortuosity exist, dependencies between tortuosity and parameters of cardiovascular system are not fully explored. In this paper we studied whether different tortuosity estimation algorithms can detect the change of blood pressure in the cylindrical segment of the vessel modeled using finite elements method. In addition we studied how does one inhomogeneity added inside the blood vessel influence the tortuosity and what are the relationships between the different tortuosity estimates and blood pressure? We found that even single inhomogeneity of the vessel wall triggers the increase of tortuosity when inner blood pressure increases. The resulting dependencies among different tortuosity estimates and blood pressure are mostly nonlinear.

Ranking of color space components for detection of blood vessels in eye fundus images

Most of the studies so far used the green component of RGB for blood vessel recognition in eye fundus images with some using luminance. However, no study was performed to justify in quantitative manner these choices. We aimed to investigate which color spaces (RGB, YIQ, HSV, HSL, XYZ’) can emphasize vasculature and vascular reflex in order to enable the usage of computationally and conceptionally simpler blood vessel detection methods.

The use of information technologies for diagnosis in ophthalmology

In 2003, a health IT programme for clinical decision support started in Lithuania. An initial goal was to create databases for ophthalmology images and to develop processing algorithms to extract diagnostically valuable information from images. We have investigated how vectors, consisting of the parameters derived from fundus images, are distributed and whether they form specific groups. When analysing the multidimensional patient data vectors, comprising all the 27 image parameters, it was impossible to separate the healthy eyes from the diseased ones. However, it was possible to simplify the system by eliminating redundant parameters and introducing new ones that represent a subset of parameters from the initial group. Thus it may prove possible to identify glaucoma using this system of parameters.

Network based clinical decision support system

Developed prototype of network based clinical decision support system consists of database of clinical data and web-based applications for signal and image analysis methods and algorithms. The methods for eye fundus image analysis and ECG P-wave morphology evaluation are the first methods covering two clinical specialties - cardiology and ophthalmology in the system. Network based database and combined analysis of the parameters obtained by means of implemented methods gives a possibility of holistic approach in clinical decision support.

Parameterization of the Optic Nerve Disk in Eye Fundus Images

Algorithms for interactive and automatic optic nerve disk parameterization are presented in this paper. Parameterization of the optic nerve disk is done by detection of the disk boundaries and calculation of parameters describing it. Automatic and interactive algorithms and user interface are designed and implemented in Matlab (The MathWorks, Inc.) programming environment. Interactive identification of the optic nerve disk boundaries was done by experienced ophthalmologist. Reference points interactively placed by physician on the nerve disk boundary then approximated by ellipse. Those boundaries were used as reference boundaries for the evaluation of automatic boundaries detection algorithm. Automatic algorithm is based on removing of vessels by segmentation and approximation and then using derivative image and verification algorithm to determine boundary location. Optic nerve disk boundary is then approximated by ellipse. Comparison of interactive and automatic algorithms shows that automatic algorithm worked well enough on more than 80% of images (in 300 images dataset).

Modification of drusen border detection algorithm based on flooding simulation

Drusen are white or yellow spots in eye fundus that consist of extracellular material. They are a sign of age-related macular degeneration (AMD) - the main cause of blindness in the developed countries and a third main cause of blindness in the whole world. Number and area of drusen are used to grade AMD. However, calculating them manually is time consuming. Thus various methods have been used to detect drusen automatically or semi-automatically. In this paper we propose an alternative method to find the borders of the drusen.

Algorithms and Results of Eye Tissues Differentiation Based on RF Ultrasound

Algorithms and software were developed for analysis of B-scan ultrasonic signals acquired from commercial diagnostic ultrasound system. The algorithms process raw ultrasonic signals in backscattered spectrum domain, which is obtained using two time-frequency methods: short-time Fourier and Hilbert-Huang transformations. The signals from selected regions of eye tissues are characterized by parameters: B-scan envelope amplitude, approximated spectral slope, approximated spectral intercept, mean instantaneous frequency, mean instantaneous bandwidth, and parameters of Nakagami distribution characterizing Hilbert-Huang transformation output. The backscattered ultrasound signal parameters characterizing intraocular and orbit tissues were processed by decision tree data mining algorithm. The pilot trial proved that applied methods are able to correctly classify signals from corpus vitreum blood, extraocular muscle, and orbit tissues. In 26 cases of ocular tissues classification, one error occurred, when tissues were classified into classes of corpus vitreum blood, extraocular muscle, and orbit tissue. In this pilot classification parameters of spectral intercept and Nakagami parameter for instantaneous frequencies distribution of the 1st intrinsic mode function were found specific for corpus vitreum blood, orbit and extraocular muscle tissues. We conclude that ultrasound data should be further collected in clinical database to establish background for decision support system for ocular tissue noninvasive differentiation.

Finding Initial Points for Drusen Detection

Drusen are white or yellow spots in eye fundus that consist of extracellular material. They are a sign of age-related macular degeneration – the main cause of blindness in the developed countries and a third main cause of blindness in the whole world. The paper discusses a method to find initial points for drusen detection based on rejection of unsuitable local maximums using high-pass and median filtering.

Estimation of blurring of optic nerve disc margin

It is known that eye fundus optic nerve disc margins get blurred in papilledema or neuritis. Doctors describe this feature of optic nerve disc by using binary classification “clear margin” or “blurred margin”. We haven’t found studies about quantitative estimation of this phenomenon during our literature analysis.

Video and inertial sensors based estimation of kinematical parameters in swimming sport

The purpose of this study was to investigate if tracking of kinematical parameters (longitudinal acceleration, velocity and pitch angle) in swimming sport is feasible using data acquired with inertial sensors: accelerometer and gyroscope. Video data analysis combined “video + inertial sensor” and “inertial sensors only” methods are presented for estimation of kinematical parameters. The equation for estimation of longitudinal acceleration from inertial sensors data is derived and complementary filter for estimation of pitch angle variation is proposed. The inertial sensor based kinematical parameters are compared with video data analysis and electromagnetic tracking system derived parameters.