No description available.

nÃo hà / Imagens de fundo de olho constituem um valioso recurso para o diagnÃstico mÃdico, pois muitas vezes apresentam indicaÃÃes de doenÃas oftÃlmicas como as da retina e atà mesmo doenÃas sistÃmicas como diabetes, hipertensÃo e arteriosclerose. Esta tese trata de algoritmos de detecÃÃo de estruturas como a fÃvea, mÃcula, exsudatos e disco Ãptico (DO) em imagens de retina. Em se tratando de algoritmos de detecÃÃo da fÃvea em imagens coloridas de retina, propomos um algoritmo assim como conjunto de regras para avaliaÃÃo dos mesmos. A detecÃÃo automÃtica desta estrutura anatÃmica à um prÃ-requisito para o diagnÃstico auxiliado por computador de vÃrias doenÃas da retina, como a degeneraÃÃo macular. Entretanto, as pequenas dimensÃes e baixo contraste da fÃvea dificultam a execuÃÃo desta tarefa de detecÃÃo. O algoritmo proposto determina a regiÃo de interesse levando em consideraÃÃo as coordenadas do DO e o fato da fÃvea ser uma Ãrea escura, homogÃnea e sem presenÃa de vasos sanguÃneos. Em seguida, o mÃtodo realiza a etapa de segmentaÃÃo dos vasos e pesquisa pela janela com menor mÃdia de intensidade de cor na imagem resultante da fusÃo entre os canais vermelho e verde. Os testes do algoritmo de detecÃÃo da fÃvea foram realizados em trÃs bases de imagens pÃblicas de referÃncia ARIA, DRIVE e MESSIDOR. Neste trabalho, propomos ainda um algoritmo de detecÃÃo de exsudatos em imagens de retina. A metodologia proposta combina agrupamento nebuloso e tÃcnicas de morfologia matemÃtica. Os resultados confirmam a melhoria no desempenho do mÃtodo de detecÃÃo quando comparado aos mÃtodos disponÃveis na literatura. Portanto, comparamos os resultados de seis algoritmos automÃticos de detecÃÃo do DO disponÃveis na literatura, utilizando dados de referÃncia das bases pÃblicas ARIA, STARE, DRIVE e MESSIDOR. O objetivo era determinar a robustez dos mesmos em detectar o DO em imagens de retina saudÃveis e com a presenÃa de patologias. Observamos que em geral os mÃtodos de detecÃÃo de DO que apresentam melhor desempenho o fazem em bases menos desafiadoras como as duas Ãltimas, ou seja, eles alcanÃam as maiores taxas de acerto. / Fundus images are valuable resource in diagnosis because they often present indications about retinal, ophthalmic, and even systemic diseases such as diabetes, hypertension, and arteriosclerosis. This thesis focuses on algorithms to detect fovea, exudates and optic disk (OD) in retina images. Regarding fovea detection algorithms in colored retina images, we propose an algorithm and furthermore a set of rules to assess them. Automatic detection of this anatomical structure is a prerequisite for computer-aided diagnosis of several retinal diseases, such as macular degeneration. However, the small dimension and weak contrast of the fovea area on retina images make difficult this task detection, directly. The proposed algorithm determines a region of interest taking into account OD coordinates and the fact that the fovea is a homogeneous dark area without blood vessels. Then, the method performs the vessel segmentation step and searches for the lowest mean color intensity window in the image that results from the fusion between the red and green channels. Tests were carried out on three public benchmark databases. In addition, this thesis proposes an algorithm for exudate detection in retina images. The proposed methodology combines fuzzy clustering and mathematical morphology techniques. The results confirm the performance improvement provided by the proposed methodology, when comparing it to other methods available in the literature. In this work, we compare the results of six different automatic algorithms for OD detection, using the public benchmark image database named ARIA, STARE, DRIVE and MESSIDOR. We aimed to test the robustness of the algorithms in detecting the OD in healthy and pathological retina images. In general, we observed that these methods performed better in less challenging databases as the two last ones, i.e. they achieved the highest success rates in DRIVE and MESSIDOR.

Ocular fundus image can provide information on pathological changes caused by local ocular diseases and early signs of certain systemic diseases. Automated analysis and interpretation of fundus images has become a necessary and important diagnostic procedure in ophthalmology. Among the features in ocular fundus image are the optic disc, fovea (central vision area), lesions, and retinal vessels. These features are useful in revealing the states of diseases in the form of measurable abnormalities such as length of diameter, change in color, and degree of tortuosity in the vessels. In addition, retinal vessels can also serve as landmarks for image-guided laser treatment of choroidal neovascularization. Thus, reliable methods for blood vessel detection that preserve various vessel measurements are needed. In this paper, we will examine the pathological issues in the analysis of retinal vessels in digital fundus images and give a survey of current image processing methods for extracting vessels in retinal images with a view to categorize them and highlight their differences and similarities. We have also implemented two major approaches using matched filter and mathematical morphology respectively and compared their performances. Some prospective research directions are identified. / Singapore-MIT Alliance (SMA)

This thesis deals with the analysis of retinal images from digital fundus camera, especially with structure of optic disc (OD). The theoretic part describes main features of the human visual system and princip of eyeground examination. The paper discussed available methods that have been used for localization of optic disc. For further work is selected proposal of the geometrical model of vessel structure by using the method of segmentation of vessels by matched filtering. Original images are tested with semi-automatic method for optic disc localization.

This thesis deals with the analysis of retinal images from digital fundus camera, especially with structure of optic disc (OD). The theoretic part describes main features of the human visual system and princip of eyeground examination. The paper discussed sevaral available methods that have been used for localization of the optic disc. In the second part is proposed the automatic detector of the optic disc based on the principle of genetic algorithm by using the method of segmentation of vessels by matched filtering. Original images are tested with this automatic method.

Retinal image analysis plays a very important role, as human gets around 90% of environment information with the help of eyes. Automation of process of retinal image analysis promotes to improve the efficiency of retinal medical examinations. The following thesis is dedicated to automatic classification methods of retinal vascular system images obtained from a digital fundus camera. Vessel classification method using classifier on the base of neural networks, which is trained and then tested on the retinal vessel segments, is investigated and implemented. In this thesis anatomical retinal survey, properties of image data from digital fundus camera and retinal image classification methods are briefly described. The last chapter is devoted to the evaluation of efficiency of retinal vessel classification with automatic methods.

This diploma thesis deals with the application of deep neural networks with focus on image segmentation. The theoretical part contains a description of deep neural networks and a summary of widely used convolutional architectures for segmentation of objects from the image. Practical part of the work was devoted to testing of an existing network architectures. For this purpose, an open-source software library Tensorflow, implemented in Python programming language, was used. A frequent problem incorporating the use of convolutional neural networks is the requirement on large amount of input data. In order to overcome this obstacle a new data set, consisting of a combination of five freely available databases was created. The selected U-net network architecture was tested by first modification of the newly created data set. Based on the test results, the chosen network architecture has been modified. By these means a new network has been created achieving better performance in comparison to the original network. The modified architecture is then trained on a newly created data set, that contains images of different types taken with various fundus cameras. As a result, the trained network is more robust and allows segmentation of retina blood vessels from images with different parameters. The modified architecture was tested on the STARE, CHASE, and HRF databases. Results were compared with published segmentation methods from literature, which are based on convolutional neural networks, as well as classical segmentation methods. The created network shows a high success rate of retina blood vessels segmentation comparable to state-of-the-art methods.

Diplomová práca je zameraná na spracovávanie obrazov sietnice s vysokým rozlíšením. Cieľom práce je zlepšiť výslednú kvalitu výsledných snímkov sietnice získaných zo sekvencie snímkov nižšej kvality. Jednotlivé snímky sú najskôr spracované pomocou bilaterálnej filtrácie a zlepšenia kontrastu. v ďalšom kroku sú odstránené rozmazané snímky a snímky zobrazujúce iné časti sietnice. Posun medzi jednotlivými snímkami v sekvencii sa odhaduje pomocou fázovej korelácie, a tieto obrazy sú potom fúzované do výsledného snímku s vysokým rozlíšením pomocou priemerovania a využitia superrozlišovacej techniky, presnejšie regularizácie pomocou bilaterálneho celkového rozptylu. Výsledné mediánové hodnoty skóre kvality získaných obrazov sú PIQUE 0.2600, NIQE 0.0701, a BRISQUE 0.3936 pre techniku priemerovania, a PIQUE 0.1063, NIQE 0.0507, and BRISQUE 0.1570 pre superrozlišovaciu techniku.

Segmentation of vessel structure is an important phase in analysis of retinal images. The resulting vessel system description may be important for diagnostic of many eye and cardiovascular diseases. A method for automatic segmentation of the vessel structure in colour retinal images is presented in the thesis. The method utilises 2D matched filtering to detect presence of short linear vessel sections of a particular thickness and orientation. The approach correlates the local image areas with a 2D masks based on a typical brightness profile perpendicular to vessels of a particular width. Three different approximated profiles are used and corresponding matched filters are designed for: thin, medium and thick vessels. The evaluation of typical vessel profiles and filter design are described in chapter 3 and chapter 4. The parametric images obtained by convolution of the image with the masks are then thresholded in order to obtain binary representation of vessel structure. The three binary representations are consequently combined to provide the best available rough vessel map, which is finalised by complementing the obviously missing vessel sections and cleaning the disconnected fractional artefacts. The thresholding algorithm and final steps of processing are mentioned in chapter 5 and chapter 6. The method has been implemented by computer and the program for automatic vessel segmentation has been developed using database of real retinal images. The efficiency of the method has been finally evaluated on images from the standard database DRIVE.

This thesis deals with method of blood vessels segmentation from retinal images acquired by fundus camera. There is explored possibility of using wavelet transform as fast outline segmentation. The thesis includes study problems of preprocessing input image and decomposition of image using 2D DWT. Furthermore there is explored possibility of parametrical images thresholding that ensue from application of 2D DWT. There are designed algorithms for cleaning off artifacts from rough vessel map of blood vessel structures. The realization of algorithm was solved in programming environment MATLAB. There was created a user control interface in graphic application GUIDE, for easy control of whole segmentation process. In conclusion of thesis is proceeded the discussion of segmentation results for images from DBME database and quantitative evaluation of results for DRIVE database images.