RESUMEN
Anterior segment optical coherence tomography (AS-OCT) is a fundamental ophthalmic imaging technique. AS-OCT images can be examined by experts and segmented to provide quantitative metrics that inform clinical decision making. Manual segmentation of these images is time-consuming and subjective, encouraging software developers in the field to automate segmentation procedures. Traditional programing segmentation approaches are being replaced by deep learning methods, which have shown state-of-the-art performance in AS-OCT image analysis. In this study, a method based on patch-based convolutional neural networks (CNN) was used to segment the three main boundaries of the cornea: the epithelium, Bowman's layer, and the endothelium. To assess the effect of the number of classes on performance, the model was designed as a patch-based boundary classifier using 4 and 8 classes. The effect of image quality was also assessed using different data distributions during the training process. While the Dice coefficient and probability revealed greater precision for the 8 class models, the boundary error metric indicated comparable performance. Additionally, for 8 class models, the image quality test had only a small negative effect on performance, which may be an indication of the robustness of the model and could also suggest that the data augmentation methods did not show significant improvement. These findings contribute to the development of automatic segmentation techniques for AS-OCT images, since patch-based methods have been largely unexplored in favor of other deep learning techniques. The overall performance of the proposed method is comparable to other well-established segmentation methods.
Asunto(s)
Redes Neurales de la Computación , Tomografía de Coherencia Óptica , Tomografía de Coherencia Óptica/métodos , Procesamiento de Imagen Asistido por Computador/métodos , Programas Informáticos , Córnea/diagnóstico por imagenRESUMEN
BACKGROUND: Anterior segment optical coherence tomography (AS-OCT) constitutes an important imaging modality to examine the anterior eye, which is commonly used in research and clinical practice. Since its introduction, a range of image analysis methods have been developed to quantify these images using different analysis techniques for various applications. This systematic review aims to provide an in-depth summary and to classify image analysis techniques found in the literature applied to AS-OCT images. METHODS: Scopus and Engineering Village databases were searched to retrieve relevant studies up to and including January 2022. Customized search statements were used along with cross reference and hand search techniques to ensure a complete coverage. Performance metrics were extracted, analyzed, and compared (when possible). RESULTS: Three main application categories were identified: glaucoma assessment, corneal segmentation, and anterior segment biometry. These three categories constitute 66% of the total studies reported in this review. Studies were also analyzed by year of publication, and since 2019 deep learning methods were favored over traditional programming or machine learning methodologies. Overall, the AS-OCT image analysis field is less developed compared to posterior segment OCT imaging. CONCLUSION: This review presents the state of the art in the field of AS-OCT image analysis. It highlights the opportunities for future areas of research, such as the expansion of DL methods and the extension to specific clinical areas that have received limited attention including surgical monitoring, contact lenses, and specific clinical conditions such as keratoconus and corneal lesions.
Asunto(s)
Queratocono , Tomografía de Coherencia Óptica , Segmento Anterior del Ojo/diagnóstico por imagen , Segmento Anterior del Ojo/patología , Biometría/métodos , Córnea , Humanos , Tomografía de Coherencia Óptica/métodosRESUMEN
PURPOSE: To analyse superficial peripapillary vascularization in non-arteritic anterior ischaemic optic neuropathy (NAION) at acute and atrophic (3 months) stage. PROCEDURES: Prospective case-control study including 6 patients with NAION and 10 age-matched healthy controls evaluated with optical coherence tomography angiography (OCT-A; Angioplex-Cirrus) at acute and atrophic stage. Apart from the -commercially provided measurements for vessel density (VD) and perfusion density (PD), a custom image analysis was used to quantify the peripapillary capillary density (PCD). RESULTS: NAION-group demonstrated a significant decrease in the PCD, VD and PD compared with fellow unaffected and control groups at acute and atrophic stage. At 3 months, the average and the temporal sector in PCD correlated with logMAR VA (-0.943, p = 0.005 and -0.829, p = 0.042 for average and temporal sectors respectively) and with MD (0.943, p = 0.005; and 0.899; p = 0.015, respectively). Over 3 months, there was a significant PCD reduction at the temporal sector and at the inner circle in VD and PD, which correlated with ganglion cell-inner plexiform layer (GCIPL) thinning over the 3 months period after the acute NAION (0.749, p = 0.020; 0.885, p = 0.002; 0.767, p = 0.016 respectively). CONCLUSION: Both strategies demonstrated a significant peripapillary microvascular dropout in NAION, but the customized analysis detected them -earlier. A progressive vessel reduction occurs within the first 3 months, which correlates with GCIPL thinning.