Academic literature on the topic 'Optic Disc Detection'

Optic disc and optic cup are one of the most recognized retinal landmarks, and there are numerous methods for their automatic detection. Segmented optic disc and optic cup are useful in providing the contextual information about the retinal image that can aid in the detection of other retinal features, but it is also useful in the automatic detection and monitoring of glaucoma. This paper proposes a deep learning based approach for the automatic optic disc and optic cup semantic segmentation, but also the new model for possible glaucoma detection. The proposed method was trained on DRIVE and DIARETDB1 image datasets and evaluated on MESSIDOR dataset, where it achieved the average accuracy of 97.3% of optic disc and 88.1% of optic cup. Detection rate of glaucoma diesis is 96.75%

With the advances in the computer field, methods and techniques in automatic image processing and analysis provide the opportunity to detect automatically the change and degeneration in retinal images. Localization of the optic disc is extremely important for determining the hard exudate lesions or neovascularization, which is the later phase of diabetic retinopathy, in computer aided eye disease diagnosis systems. Whereas optic disc detection is fairly an easy process in normal retinal images, detecting this region in the retinal image which is diabetic retinopathy disease may be difficult. Sometimes information related to optic disc and hard exudate information may be the same in terms of machine learning. We presented a novel approach for efficient and accurate localization of optic disc in retinal images having noise and other lesions. This approach is comprised of five main steps which are image processing, keypoint extraction, texture analysis, visual dictionary, and classifier techniques. We tested our proposed technique on 3 public datasets and obtained quantitative results. Experimental results show that an average optic disc detection accuracy of 94.38%, 95.00%, and 90.00% is achieved, respectively, on the following public datasets: DIARETDB1, DRIVE, and ROC.

Glaucoma is the second leading cause of loss of vision in the world. Examining the head of optic nerve (cup-to-disc ratio) is very important for diagnosing glaucoma and for patient monitoring after diagnosis. Images of optic disc and optic cup are acquired by fundus camera as well as Optical Coherence Tomography. The optic disc and optic cup segmentation techniques are used to isolate the relevant parts of the retinal image and to calculate the cup-to-disc ratio. The main objective of this paper is to review segmentation methodologies and techniques for the disc and cup boundaries which are utilized to calculate the disc and cup geometrical parameters automatically and accurately to help the professionals in the glaucoma to have a wide view and more details about the optic nerve head structure using retinal fundus images. We provide a brief description of each technique, highlighting its classification and performance metrics. The current and future research directions are summarized and discussed.

Automated retinal image analysis has been emerging as an important diagnostic tool for early detection of eye-related diseases such as glaucoma and diabetic retinopathy. In this paper, we have presented a robust methodology for optic disc detection and boundary segmentation, which can be seen as the preliminary step in the development of a computer-assisted diagnostic system for glaucoma in retinal images. The proposed method is based on morphological operations, the Circular Hough transform and the Grow Cut algorithm. The morphological operators are used to enhance the optic disc and remove the retinal vasculature and other pathologies. The optic disc center is approximated using the Circular Hough transform, and the Grow Cut algorithm is employed to precisely segment the optic disc boundary. The method is quantitatively evaluated on five publicly available retinal image databases DRIVE, DIARETDB1, CHASE_DB1, DRIONS-DB, Messidor and one local Shifa Hospital Database. The method achieves an optic disc detection success rate of 100% for these databases with the exception of 99.09% and 99.25% for the DRIONS-DB, Messidor, and ONHSD databases, respectively. The optic disc boundary detection achieved an average spatial overlap of 78.6%, 85.12%, 83.23%, 85.1%, 87.93%, 80.1%, and 86.1%, respectively, for these databases. This unique method has shown significant improvement over existing methods in terms of detection and boundary extraction of the optic disc.

<p>The rapid growth of technology makes it possible to implement in immediate diagnosis for patients using image processing. By using morphological processing and adaptive thresholding method for segmentation of optic disc and optic cup, various sizes of retinal fundus images captured through fundus camera from online databases can be processed. This paper explains the use of color channel separation method for pre-processing to remove noise for better optic disc and optic cup segmentation. Noise removal will improve image quality and in return help to increase segmentation standard. Then, morphological processing and adaptive thresholding method is used to extract out optic disc and optic cup from fundus image. The proposed method is tested on two publicly available online databases: RIM-ONE and DRIONS-DB. On RIM-ONE database, the average PSNR value acquired is 0.01891 and MSE is 65.62625. Meanwhile, for DRIONS-DB database, the best PSNR is 64.0928 and the MSE is 0.02647. In conclusion, the proposed method can successfully filter out any unwanted noise in the image and are able to help clearer optic disc and optic cup segmentation to be performed.</p>

This paper proposes a method for the automatic detection of optic disc in retinal images. In the diagnosis and grading, the essential step is recognition of optic disk for diabetic retinopathy. The analysis of directional cross section profile focused on the local maximum pixel of pre-processed image is realized by the proposed method using optic disc detection. Each profile is implemented by peak detection and property like shape, size and height of the peak are estimated. The statistical measure of the estimated values for the attributes, where the orientation of the cross-section changes the constitute feature used in morphological classification to exclude encourages candidates. The result is to find the patient is affected by diabetics or not.Keywords: Diabetic retinopathy, Optic disk, Naives Bayes algorithm, Local maximum region.

<p>Identifying the position of the optic disc on the retinal fundus image is a technique that is often used in medical diagnosis, treatment and monitoring processes. Determination of the intensity of the bright colors that belongs to the optic disc on a normal retinal image by the help of image processing algorithms is a fairly easy process. However, determining the optic disc on a retinal image including the diabetic retinopathy disease is a more difficult process. The reason for this difficulty is the existence of many regions that have the same light intensity in different parts of the retina. In this study, a new method for supplying the automatic determination of the optic disc in a recursive manner is proposed. By the help of OpenCV library, automatic determination process of the optic disc on the retinal fundus images including the diabetic retinopathy disease, has been implemented. Circular regions with maximum brightness values in the retinal images that were normalized and passed through the denoising process were determined and these regions were analyzed if they are optic disc or not. This process basically consists of two steps: In the first step, the possible optic disc candidate regions were determined recursively and in the second step, by the help of Gabor filter kernels, these regions were analyzed and it’s provided to decide if they are optic disc or not. This study is based on a dataset that has 89 images including diabetic retinopathy disease. Performance of this system is tested on these images and also on the images that the red, green, blue color channels and Contrast Limited Adaptive Histogram Equalization (CLAHE) retinas were obtained. Most accurate determination of the position of the optic disc is obtained with retinas, implemented process CLAHE, including the best success rate of 89.88%.</p><p> </p>Keywords: Optic disc, diabetic retinopathy, recursively, circular region, gabor filter kernels.

The optic disc is a key anatomical structure in retinal images. The ability to detect optic discs in retinal images plays an important role in automated screening systems. Inspired by the fact that humans can find optic discs in retinal images by observing some local features, we propose a local feature spectrum analysis (LFSA) that eliminates the influence caused by the variable spatial positions of local features. In LFSA, a dictionary of local features is used to reconstruct new optic disc candidate images, and the utilization frequencies of every atom in the dictionary are considered as a type of “spectrum” that can be used for classification. We also employ the sparse dictionary selection approach to construct a compact and representative dictionary. Unlike previous approaches, LFSA does not require the segmentation of vessels, and its method of considering the varying information in the retinal images is both simple and robust, making it well-suited for automated screening systems. Experimental results on the largest publicly available dataset indicate the effectiveness of our proposed approach.

Automated localization of the optic disc and fovea are important in the field of analysis of fundus images. We introduce a simultaneous detection method for optic disc and fovea with an enhancement and correction step. In the first step, a set of features are extracted from the color fundus image, and the relationship between the feature set and a distance variable d is established during training phase. For a test image, the same set of features is measured and the distance to the optic disc and fovea can be estimated using k-nearest-neighbor regression. A probability image is generated during this step. In the second, a second k-nearest-neighbor regression is applied on the probability image. Detected high likelihood regions from the first step can be enhanced only if they satisfy the trained relationship. The detected regions that do not get support from the other detected structure will be suppressed. 150 color fundus images were used to train the system. 50 color fundus images were used to test the system. The distance error for the optic disc is 9.8±8.3 pixels. The distance error for the fovea is 13.7±6.6 pixels.

This work is focused on automatic detection of optic disc in retinal images. There is briefly described anatomy of human eye, principles of retinal imaging and also overview of the methods used for optic disc detection. The practical part describes developed procedures for optic disc detection, ie detection based on watershed transform, active contours and also on region growing technique. The main method of this work is the method of circular transformation, which as the only one allowed to detect the optic disc on the images of video ophtalmoscope and also on the high quality images from fundus cameras. This method was tested on three datasets. The average overlap 92,44 % was achieved for HRF dataset, 91,03 for DRIONS dataset and 77,36 for images of video ophtalmoscope.

The contributions of this work are in the development of new and state of the art algorithms for retinal image analysis including optic disc detection, tortuosity estimation, and cross-over abnormality detection. The retina is one of the only areas of the human body that blood vessels can be visualized noninvasively. Retinal imaging has become a standard in the ophthalmologist’s office because it is an easy and inexpensive way to monitor not just eye health, but also systemic vascular diseases. Changes to the retinal vasculature can be the early signs of diseases such as diabetic and hypertensive retinopathy, of which early detection can save vision, money, and improve overall health for the patient. When looking at the retinal vasculature, ophthalmologists generally rely on a qualitative assessment which can make comparisons over time or between different ophthalmologists difficult. Computer aided systems are now able to quantify what the ophthalmologist is qualitatively measuring in what they consider to be the most important features of the vasculature. These include, but are not limited to, tortuosity, arteriolar narrowing, cross-over abnormalities, and artery-vein (AV) ratio. The University of Padova has created a semi-automatic system for detecting and quantifying retinal vessels starting from optic disc detection, vessel segmentation, width estimation, tortuosity calculation, AV classification, and AV ratio. We propose a new method for optic disc detection that converts the retinal image into a graph and exploits vessel enhancement methods to calculate edge weights in finding the shortest path between pairs of points on the periphery of the image. The line segment with the maximum number of shortest paths is considered the optic disc location. The method was tested on three publicly available datasets: DRIVE, DIARETDB1, and Messidor consisting of 40, 89, and 1200 images and achieved an accuracy of 100, 98.88, and 99.42% respectively. The second contribution is a new algorithm for calculating abnormalities at AV crossing points. In retinal images, Gunn’s sign appears as a tapering of the vein at a crossing point, while Salus’s sign presents as an S-shaped curving. This work presents a method for the automatic quantification of these two signs once a crossover has been detected; combining segmentation, artery vein classification, and morphological feature extraction techniques to calculate vein widths and angles entering and exiting the crossover. Results on two datasets show separation between the two classes and that we can reliably detect and quantify these signs under the right conditions. The last contribution in tortuosity consists of two parts. A comparative study was performed on several of the most popular methods for tortuosity estimation on a new vessel dataset. Results show that several methods have good Cohen’s kappa agreement with both graders, while the tortuosity density metric has the highest single metric average agreement across vessel type and grader. The second is a new way to enhance curvature in segmented vessels based on a difference of Gabor filters to create a curvature enhanced image. The proposed method was tested on the RET-TORT database using several methods to calculate tortuosity, and had best Pearson’s correlation of .94 for arteries and .882 for veins, outperforming single mathematical formulations on the data. This held true after testing the method on the propose dataset as well, having higher correlation values across grader and vessel type compared with other tortuosity metrics. Summary of Results: The optic disc detection method was tested on three publicly available datasets: DRIVE, DIARETDB1, and Messidor consisting of 40, 89, and 1200 images and achieved an accuracy of 100, 98.88, and 99.42% respectively. The AV nicking quantification method was tested on a small dataset of 10 crossing provided by doctors at Papageorgiou Hospital, Aristotle University of Thessaloniki, Thessaloniki, Greece. Results showed separation between the normal and abnormal classes for both the Gunn and Salus sign. The method was then tested on a larger, publicly available dataset which showed good separation for the Gunn sign. The proposed tortuosity method was tested on the RET-TORT database using several methods to calculate tortuosity, and had best Pearson’s correlation of .94 for arteries and .882 for veins, outperforming single mathematical formulations on the data. It was then tested on the dataset proposed in this thesis, further corroborating the effectiveness of the method.<br>I contributi di questo lavoro sono per lo sviluppo di nuovi e lo stato degli algoritmi d'arte per l'analisi di immagini tra cui il rilevamento della retina ottica del disco, la stima tortuosità, e anomalia rilevamento cross-over. La retina è una delle poche zone del corpo umano che vasi sanguigni possono essere visualizzate in maniera non invasiva. imaging della retina è diventato uno standard nell'ufficio del oculista Poiché si tratta di un modo semplice e poco costoso per monitorare non solo la salute degli occhi, ma anche le malattie vascolari sistemiche. Le modifiche al sistema vascolare della retina possono essere i primi segni di malattie come la retinopatia diabetica e ipertensiva, di cui la diagnosi precoce può salvare la visione, il denaro, e migliorare la salute generale del paziente. Se si guarda alla vascolarizzazione della retina, oftalmologi in genere si basano su una valutazione qualitativa che può fare comparazioni nel tempo o tra i diversi oculisti difficile. Computer Aided sistemi sono ora incendio quantificare ciò che l'oculista è qualitativamente misura in quello che considerano essere le caratteristiche più importanti del sistema vascolare. Questi includono, ma non sono limitati a, tortuosità, arteriolare restringimento, anomalie di crossover, e il rapporto arteria-vena (AV). L'Università di Padova ha creato un sistema semi-automatico per la rilevazione e quantificazione vasi retinici a partire dalla rilevazione ottica del disco, la segmentazione nave, la stima di larghezza, il calcolo tortuosità, la classificazione AV, e il rapporto di AV. Abbiamo proposto un nuovo metodo per il rilevamento ottico che converte l'immagine retinica in un grafico e sfrutta disco metodi di aumento del vaso per calcolare i pesi bordo nel trovare il percorso più breve tra coppie di punti sulla periferia dell'immagine. Il segmento linea con il numero massimo di percorsi più brevi è considerata la posizione del disco ottico. Il metodo è stato testato su tre insiemi di dati accessibili al pubblico: DRIVE, DIARETDB1, e Messidor Composto da 40, 89, e 1200 immagini e ha raggiunto una precisione di 100, 98.88, e 99.42% respectively. Il secondo contributo è un nuovo algoritmo di calcolo anomalie AV ai punti di attraversamento. Nelle immagini della retina, segno di Gunn appare come una rastremazione della vena in un punto di passaggio, mentre il segno di Salus presenta come una curva a forma di S. Questo lavoro presenta un metodo per la quantificazione automatica di questi due segni once a incrocio è stato rilevato; combinando la segmentazione, l'arteria di classificazione della vena, e le tecniche di estrazione delle caratteristiche morfologiche per calcolare le larghezze delle vene e gli angoli che entrano ed escono il crossover. Risultati su due serie di dati mostrano la separazione tra le due classi e che possiamo in modo affidabile rilevare e quantificare Questi segni sotto le giuste condizioni. L'ultimo contributo in tortuosità compone di due parti. Uno studio comparativo è stato condotto su alcuni dei metodi più diffusi per la stima su un nuovo insieme di dati nave tortuosità. Che i risultati mostrano diversi metodi avere buon accordo con Cohen kappa Entrambi i selezionatori, mentre la metrica densità di tortuosità ha la più alta accordo metrica singola media di tipo di nave e selezionatore. Il secondo è un nuovo modo per migliorare curvature nei vasi segmentati sulla base di una differenza di Gabor filtri per creare una curvatura immagine migliorata. Theproposed Il metodo è stato testato su database RET-TORT utilizzando diversi metodi per calcolare tortuosità, e aveva più di correlazione di Pearson di .94 per arterie e vene per 0,882, superando singole formulazioni matematiche alla data. Questo valeva dopo aver testato il metodo proposto sul set di dati e, avendo valori di correlazione più elevati in tutta grader e tipo di imbarcazione tortuosità Rispetto ad altre metriche. Sintesi dei risultati: Il metodo di rilevazione disco ottico è stato testato su tre insiemi di dati accessibili al pubblico: DRIVE, DIARETDB1, e Messidor Composto da 40, 89, e 1200 immagini e ha raggiunto una precisione di 100, 98.88, e 99.42% respectively. Il metodo di quantificazione AV intaccare è stato testato su un piccolo insieme di dati di 10 attraversamento forniti dai medici Papageorgiou Hospital, Università Aristotele di Salonicco, Salonicco, Grecia. I risultati hanno mostrato separazione tra le normali e anormali Entrambe le classi per il segno Gunn e Salus. Il metodo è stato poi testato su un set di dati più grandi, a disposizione del pubblico che ha mostrato una buona separazione per il segno Gunn. Il metodo tortuosità Theproposed è stato testato su database RET-TORT utilizzando diversi metodi per calcolare tortuosità, e aveva più di correlazione di Pearson di .94 per arterie e vene per 0,882, superando singole formulazioni matematiche alla data. E 'stato poi testato su set di dati Theproposed in questa tesi, Ulteriori confermano l'efficacia del metodo.

Dissertação de mestrado em Engenharia de Informática<br>The Fovea and Optic Disc are relevant anatomical eye structures to diagnose various diseases. Its automatic detection can provide both a cost reduction when analysing large populations and improve the effectiveness of ophthalmologists and optometrists. This dissertation describes a methodology to automatically detect these structures and analyses a, CPU only, MATLAB implementation of this methodology. RetScan is a port to a freeware environment of this methodology, its functionality and performance are evaluated and compared to the original. The results of both evaluations lead to a discussion on possible improvements in the metodology that influence the functionality and performance. The resulting improvements are implemented and integrated in RetScan. To further improve performance, a parallelization of RetScan to take advantage of a multi-core architecture or a CUDA-enabled accelerator was designed, coded and evaluated.This evaluation reveals that RetScan achieves its best throughput efficiency using a multi-core architecture only and analysing several images at once. For one image usage, using multi-core only is also the best solution, but with a small speed-up. The usage of CUDA-enabled accelerators is not recommended for this scope as the images are small and the cost of the data transfer to and from the accelerator has a severe impact on performance.<br>A Fóvea e o Disco Ótico são estruturas oculares importantes quando se procura diagnosticar doenças no olho. A sua deteção automática permite reduzir o custo de um rastreio a grandes populações e também aumentar a eficácia de oftalmologistas e optometristas. Nesta dissertação é descrita uma metodologia para detetar estas estruturas automaticamente e é analisada uma implementação em MATLAB desta metodologia. RetScan é o resultado do porte para um ambiente de desenvolvimento com ferramentas livres (open source) desta metodologia. O RetScan é avaliado quer em funcionalidade, quer em performance. Os resultados da avaliação levam a uma reflexão sobre mudanças a realizar à metodologia para melhorar os resultados em ambas as avaliações. Estas melhorias são implementadas e integradas no RetScan. Para melhorar a sua performance é também realizada um paralelização do RetScan de forma a que tire partido de uma arquitetura multi-core ou de um acelerador compatível com CUDA. Após realizar uma nova avaliação conclui-se que o RetScan atinge o seu melhor débito de dados (throughput) quando usa apenas os CPUs numa arquitetura multi-core e analisando várias imagens em paralelo. Para a análise de uma só imagem, o uso apenas de CPUs numa arquitetura multi-core também é o melhor resultado, embora tenha um ganho (speed up) reduzido. O uso de aceleradores compatíveis com CUDA não é recomendado neste âmbito pois as imagens têm um tamanho reduzido e o custo da transferência de e para estes aceleradores tem um grande impacto no tempo total

With the development of digital image techniques, modeling, and processing, automatic detection of diseases like diabetic retinopathy, Diabetic Macular Edema and glaucoma etc. has become an important screening task. Detection of optic disc pixel and its boundary are main and essential tasks concerning the retinal images to automate the calculation related to some prevailing systemic disorders, such as hypertension, cardiovascular diseases, diabetes and Glaucoma etc. Extraction of Optic Disc should be done properly so as to correctly identify these diseases. This report presents a new method which tries to detect Optic Disc more correctly using Morphological techniques followed by edge detection and Circular Hough Transform. For this, it first pre-processing is done for quality enhancement. Next step is to detects Optic Disc Pixel. Then blood vessels are extracted by Matching Filter-First Order Gaussian Derivatives. Lastly Optic Disc boundary localization is done using circular Hough Transform. For this purpose, Dataset of DIARETDB1 is used consisting of 89 images. The result of the proposed method has an average overlapping of 0.82 and Optic Disc localization accuracy of 88.75%. Result of Proposed method is then gets compared with other existed methods in the Result and Analysis section.

With the development of digital image techniques, modeling, and processing, automatic detection of diseases like diabetic retinopathy, Diabetic Macular Edema and glaucoma etc. has become an important screening task. Detection of optic disc pixel and its boundary are main and essential tasks concerning the retinal images to automate the calculation related to some prevailing systemic disorders, such as hypertension, cardiovascular diseases, diabetes and Glaucoma etc. Extraction of Optic Disc should be done properly so as to correctly identify these diseases. This report presents a new method which tries to detect Optic Disc more correctly using Morphological techniques followed by edge detection and Circular Hough Transform. For this, it first pre-processing is done for quality enhancement. Next step is todetects OpticDisc Pixel. Then blood vessel sare extracted by Matching Filter-First Order Gaussian Derivatives. Lastly Optic Disc boundary localization is done using circular Hough Transform. For this purpose, Dataset of DIARETDB1 is used consisting of 89 images. The result of the proposed method has an average overlapping of 0.82 and Optic Disc localization accuracy of 88.75%. Result of Proposed method is then gets compared with other existed methods in the Result and Analysis section.

Approximate Nearest-Neighbour Field (ANNF\ maps between two related images are commonly used by computer vision and graphics community for image editing, completion, retargetting and denoising. In this work we generalize ANNF computation to unrelated image pairs. For accurate ANNF map computation we propose Feature Match, in which the low-dimensional features approximate image patches along with global colour adaptation. Unlike existing approaches, the proposed algorithm does not assume any relation between image pairs and thus generalises ANNF maps to any unrelated image pairs. This generalization enables ANNF approach to handle a wider range of vision applications more efficiently. The following is a brief description of the applications developed using the proposed Feature Match framework. The first application addresses the problem of detecting the optic disk from retinal images. The combination of ANNF maps and salient properties of optic disks leads to an efficient optic disk detector that does not require tedious training or parameter tuning. The proposed approach is evaluated on many publicly available datasets and an average detection accuracy of 99% is achieved with computation time of 0.2s per image. The second application aims to super-resolve a given synthetic image using a single source image as dictionary, avoiding the expensive training involved in conventional approaches. In the third application, we make use of ANNF maps to accurately propagate labels across video for segmenting video objects. The proposed approach outperforms the state-of-the-art on the widely used benchmark SegTrack dataset. In the fourth application, ANNF maps obtained between two consecutive frames of video are enhanced for estimating sub-pixel accurate optical flow, a critical step in many vision applications. Finally a summary of the framework for various possible applications like image encryption, scene segmentation etc. is provided.

Approximate Nearest-Neighbour Field (ANNF\ maps between two related images are commonly used by computer vision and graphics community for image editing, completion, retargetting and denoising. In this work we generalize ANNF computation to unrelated image pairs. For accurate ANNF map computation we propose Feature Match, in which the low-dimensional features approximate image patches along with global colour adaptation. Unlike existing approaches, the proposed algorithm does not assume any relation between image pairs and thus generalises ANNF maps to any unrelated image pairs. This generalization enables ANNF approach to handle a wider range of vision applications more efficiently. The following is a brief description of the applications developed using the proposed Feature Match framework. The first application addresses the problem of detecting the optic disk from retinal images. The combination of ANNF maps and salient properties of optic disks leads to an efficient optic disk detector that does not require tedious training or parameter tuning. The proposed approach is evaluated on many publicly available datasets and an average detection accuracy of 99% is achieved with computation time of 0.2s per image. The second application aims to super-resolve a given synthetic image using a single source image as dictionary, avoiding the expensive training involved in conventional approaches. In the third application, we make use of ANNF maps to accurately propagate labels across video for segmenting video objects. The proposed approach outperforms the state-of-the-art on the widely used benchmark SegTrack dataset. In the fourth application, ANNF maps obtained between two consecutive frames of video are enhanced for estimating sub-pixel accurate optical flow, a critical step in many vision applications. Finally a summary of the framework for various possible applications like image encryption, scene segmentation etc. is provided.

This work, integrated in Critical Health, presents a new process to detect fovea and optic disk in retinal images and the application of some technologies to do the automation of Key Performance Indicators process (KPI). The proposed method consists in ve steps: selection of an area in the image where the optic disk is located using Sobel operator, extraction of optic disk boundaries applying the Hough transform to detect center and diameter of optic disk, detection of the ROI (region of interest) where the fovea is located based on the optic disk center and its diameter and detection of the fovea within the ROI. The developed algorithm has been tested in a proprietary dataset with 1464 images (with ground truth generated by experts) and with some public datasets. Retmarker is an image processing product developed by Critical Health. The KPI is a process implemented in Critical Health to test Retmarker with the goal to reach the optimal performance. This process is currently highly manual and performed on a weekly basis, demanding a considerable amount of man-hours per year. A new plan was implemented to make this process fully automated. Keywords: Biomedical image processing, Digital images, Filtering, Image segmentation, Anatomical structure, Automation, Database systems.