Vetnuus | October 2025 27 These confirmations highlight the effectiveness of advanced deep learning techniques in diagnosing canine eye diseases from images captured in real-world settings. The key contributions of this research are: An image dataset for machine learning of canine eye diseases: We developed and published the DogEyeSeg4 dataset, which includes a diverse set of real-world clinical images of canine eyes. This dataset reflects variations in lighting, breed, and image quality, providing a robust foundation for training and evaluating machine learning models in unconstrained environments. Deep convolutional neural network model for recognition of canine eye clinical symptoms and diseases from still images in unconstrained environments: We successfully implemented and tested convolutional neural network models that accurately detect and recognise canine ocular symptoms and diseases in images captured under real-world conditions. The models demonstrated high accuracy in localising the eye region and identifying key ocular symptoms, even in instances such as variable lighting and patient movement. Deep neural network based on U-Net for segmentation of canine eye clinical symptoms from still images in unconstrained environments: By enhancing the standard U-Net architecture with advanced backbone networks like ResNet34 and employing combined loss functions (Dice and Focal Loss), and augmented input, we developed the U-Net(RSD) model. This model achieved superior performance in segmenting key ocular symptoms, providing the precise delineation necessary for accurate diagnosis. An improved method for segmentation of canine eye conditions based on U-Net: We introduced modifications to the U-Net network’s input and architecture, including synthetized images and the use of a custom attention mechanism in the form of heatmaps generated from SSD eye detection. These enhancements led to improved segmentation accuracy and reduced false positives. In summary, this research demonstrates the potential of integrating advanced computer vision techniques with deep learning models to enhance the diagnosis of canine ocular diseases. The development of the DogEyeSeg4 dataset and the improved U-Netbased segmentation methods contribute valuable resources and methodologies for future research in veterinary ophthalmology. The successful application of models like U-Net(RSD) and GPT-4o showcases the feasibility of deploying such technologies in realworld clinical settings, ultimately aiming to improve diagnostic accuracy and patient outcomes in veterinary medicine. Despite these advancements, the research identified limitations such as the models’ sensitivity to image quality and difficulties in diagnosing subtle conditions like corneal ulcers. Addressing these challenges is significant for developing more accessible and reliable diagnostic tools. Future research should focus on expanding the dataset, refining model architectures, and integrating multimodal data to further enhance automated veterinary diagnostics and develop more comprehensive approaches to animal health care. v 7. REFERENCES F. Abdullah et al., “A Review on Glaucoma Disease Detection Using Computerized Techniques,” IEEE Access, vol. 9, pp. 37311–37333, 2021, doi: 10.1109/ACCESS.2021.3061451. Y. Hagiwara et al., “Computer-aided diagnosis of glaucoma using fundus images: A review,” Computer Methods and Programs in Biomedicine, vol. 165, pp. 1–12, Oct. 2018, doi: 10.1016/j.cmpb.2018.07.012. M. S. Haleem, L. Han, J. van Hemert, and B. Li, “Automatic extraction of retinal features from colour retinal images for glaucoma diagnosis: A review,” Computerized Medical Imaging and Graphics, vol. 37, no. 7–8, pp. 581–596, Oct. 2013, doi: 10.1016/j.compmedimag.2013.09.005. L. Li, M. Xu, X. Wang, L. Jiang, and H. Liu, “Attention Based Glaucoma Detection: A Large-Scale Database and CNN Model,” in 2019 IEEE/ CVF Conference on Computer Vision and Pattern Recognition (CVPR), Long Beach, CA, USA: IEEE, Jun. 2019, pp. 10563–10572. doi: 10.1109/ CVPR.2019.01082. N. Chakrabarty and S. Chatterjee, “A Novel Approach to Glaucoma Screening using Computer Vision,” in 2019 International Conference on Smart Systems and Inventive Technology (ICSSIT), Tirunelveli, India: IEEE, Nov. 2019, pp. 881–884. doi: 10.1109/ICSSIT46314.2019.8987803. Zhuo Zhang et al., “ORIGA-light: An online retinal fundus image database for glaucoma analysis and research,” in 2010 Annual International Conference of the IEEE Engineering in Medicine and Biology, Buenos Aires: IEEE, Aug. 2010, pp. 3065–3068. doi: 10.1109/ IEMBS.2010.5626137. J. Sivaswamy, S. R. Krishnadas, G. Datt Joshi, M. Jain, and A. U. Syed Tabish, “Drishti-GS: Retinal image dataset for optic nerve head(ONH) segmentation,” in 2014 IEEE 11th International Symposium on Biomedical Imaging (ISBI), Apr. 2014, pp. 53–56. doi: 10.1109/ ISBI.2014.6867807. S. Kanse and D. Yadav, “Retinal Fundus Image for Glaucoma Detection: A Review and Study,” Journal of Intelligent Systems, vol. 28, pp. 43–56, Jan. 2019, doi: 10.1515/jisys-2016-0258. F. Fumero, S. Alayón, J. L. Sanchez, J. Sigut, and M. Gonzalez-Hernandez, “RIM-ONE: An open retinal image database for optic nerve evaluation,” presented at the Int. Sym. on CBMS, Jul. 2011, pp. 1–6. doi: 10.1109/ CBMS.2011.5999143. B. Luo, J. Shen, Y. Wang, and M. Pantic, “The iBUG Eye Segmentation Dataset,” p. 9 pages, 2019, doi: 10.4230/OASICS.ICCSW.2018.7. S. J. Garbin, Y. Shen, I. Schuetz, R. Cavin, G. Hughes, and S. S. Talathi, “OpenEDS: Open Eye Dataset,” May 17, 2019, arXiv: arXiv:1905.03702. Accessed: Nov. 22, 2022. [Online]. Available: http://arxiv.org/ abs/1905.03702 H. Proenca, S. Filipe, R. Santos, J. Oliveira, and L. A. Alexandre, “The UBIRIS.v2: A Database of Visible Wavelength Iris Images Captured On-theMove and At-a-Distance,” IEEE Trans. Pattern Anal. Mach. Intell., vol. 32, no. 8, pp. 1529–1535, Aug. 2010, doi: 10.1109/TPAMI.2009.66. W. Fuhl, G. Kasneci, and E. Kasneci, “TEyeD: Over 20 million real-world eye images with Pupil, Eyelid, and Iris 2D and 3D Segmentations, 2D and 3D Landmarks, 3D Eyeball, Gaze Vector, and Eye Movement Types,” in 2021 IEEE International Symposium on Mixed and Augmented Reality (ISMAR), Oct. 2021, pp. 367–375. doi: 10.1109/ ismar52148.2021.00053. A. R. Strom, M. Hässig, T. M. Iburg, and B. M. Spiess, “Epidemiology of canine glaucoma presented to University of Zurich from 1995 to 2009. Part 1: Congenital and primary glaucoma (4 and 123 cases),” Veterinary Ophthalmology, vol. 14, no. 2, pp. 121–126, 2011, doi: https://doi.org/10.1111/j.1463-5224.2010.00855.x. D. A. J. Johnsen, D. J. Maggs, and P. H. Kass, “Evaluation of risk factors for development of secondary glaucoma in dogs: 156 cases (1999– 2004),” Journal of the American Veterinary Medical Association, vol. 229, no. 8, pp. 1270–1274, 2006, doi: 10.2460/javma.229.8.1270. M. Boevé and F. Stades, “[Glaucoma in dogs and cats. Review and retrospective evaluation of 421 patients. I. Pathobiological Leading Article
RkJQdWJsaXNoZXIy OTc5MDU=