| Abstract [eng] |
Research purpose - Create multi-modal methodology eye physical and functional properties of complex underlying data evaluation and classification of estimates generalized pathology building. Research tasks: 1. Investigate the lens opacities and fundus image color characteristics of interfaces. 2. Create eye physical and functional properties of underlying data aggregation methodologies. 3. Assess the complex multimodal information and aggregate assessment of eye diseases classification markings opportunities. Research object – Multimodal fundus images. Methodology. The study was to collect Lithuanian University of Health Sciences Kaunas Clinics Public Eye Clinic 25 patients (women and men) data. A study carried out used fundus video shooting, optical coherence tomography and retinal sensitivity test. Total survey to collect data that represent multimodal information was to combine five layers of two-dimensional set. All data have been adapted to each other, following this principle, that a simple form of pixel data represent information on the same physical point in the retina. Results: Available data according to Fisher's statistics created three canonical function, which according to Wilk's Lambda criterion two canonical functions showed the highest statistical values. The first and second canonical function is recommended for the eye disease classification. Conclusions: Examination of the eye lens turbidity, effects, observed that the opacity of the lens has a tendency to change the eye colour gamut changes in the bottom of the image. The evaluation of multimodal information components weight, noticed that the bottom of the eye image of green and red colours and retinal sensitivity and thickness are important for eye diseases classification. There is a limited amount of time available to take, is not enough to create a real clinical significance signs of disease classification system, but it is an illustration of the principle of multidimensional analysis which we hope to develop symptoms of many diseases classification systems. |
