Neuron-level explainable AI for Alzheimer's Disease assessment from fundus images.

Alzheimer's Disease (AD) is a progressive neurodegenerative disease and the leading cause of dementia. Early diagnosis is critical for patients to benefit from potential intervention and treatment. The retina has emerged as a plausible diagnostic site for AD detection owing to its anatomical connection with the brain. However, existing AI models for this purpose have yet to provide a rational explanation behind their decisions and have not been able to infer the stage of the disease's progression. Along this direction, we propose a novel model-agnostic explainable-AI framework, called Granu la ̲ r Neuron-le v ̲ el Expl a ̲ iner (LAVA), an interpretation prototype that probes into intermediate layers of the Convolutional Neural Network (CNN) models to directly assess the continuum of AD from the retinal imaging without the need for longitudinal or clinical evaluations. This innovative approach aims to validate retinal vasculature as a biomarker and diagnostic modality for evaluating Alzheimer's Disease. Leveraged UK Biobank cognitive tests and vascular morphological features demonstrate significant promise and effectiveness of LAVA in identifying AD stages across the progression continuum.

Unsupervised OCT image despeckling with ground-truth- and repeated-scanning-free features.

Optical coherence tomography (OCT) can resolve biological three-dimensional tissue structures, but it is inevitably plagued by speckle noise that degrades image quality and obscures biological structure. Recently unsupervised deep learning methods are becoming more popular in OCT despeckling but they still have to use unpaired noisy-clean images or paired noisy-noisy images. To address the above problem, we propose what we believe to be a novel unsupervised deep learning method for OCT despeckling, termed Double-free Net, which eliminates the need for ground truth data and repeated scanning by sub-sampling noisy images and synthesizing noisier images. In comparison to existing unsupervised methods, Double-free Net obtains superior denoising performance when trained on datasets comprising retinal and human tissue images without clean images. The efficacy of Double-free Net in denoising holds significant promise for diagnostic applications in retinal pathologies and enhances the accuracy of retinal layer segmentation. Results demonstrate that Double-free Net outperforms state-of-the-art methods and exhibits strong convenience and adaptability across different OCT images.

A novel vector field analysis for quantitative structure changes after macular epiretinal membrane surgery.

The aim of this study was to introduce novel vector field analysis for the quantitative measurement of retinal displacement after epiretinal membrane (ERM) removal. We developed a novel framework to measure retinal displacement from retinal fundus images as follows: (1) rigid registration of preoperative retinal fundus images in reference to postoperative retinal fundus images, (2) extraction of retinal vessel segmentation masks from these retinal fundus images, (3) non-rigid registration of preoperative vessel masks in reference to postoperative vessel masks, and (4) calculation of the transformation matrix required for non-rigid registration for each pixel. These pixel-wise vector field results were summarized according to predefined 24 sectors after standardization. We applied this framework to 20 patients who underwent ERM removal to obtain their retinal displacement vector fields between retinal fundus images taken preoperatively and at postoperative 1, 4, 10, and 22 months. The mean direction of displacement vectors was in the nasal direction. The mean standardized magnitudes of retinal displacement between preoperative and postoperative 1 month, postoperative 1 and 4, 4 and 10, and 10 and 22 months were 38.6, 14.9, 7.6, and 5.4, respectively. In conclusion, the proposed method provides a computerized, reproducible, and scalable way to analyze structural changes in the retina with a powerful visualization tool. Retinal structural changes were mostly concentrated in the early postoperative period and tended to move nasally.

Functional OCT reveals anisotropic changes of retinal flicker-evoked vasodilation.

The purpose of this study is to verify the effect of anisotropic property of retinal biomechanics on vasodilation measurement. A custom-built optical coherence tomography (OCT) was used for time-lapse imaging of flicker stimulation-evoked vessel lumen changes in mouse retinas. A comparative analysis revealed significantly larger (18.21%) lumen dilation in the axial direction compared to the lateral (10.77%) direction. The axial lumen dilation predominantly resulted from the top vessel wall movement toward the vitreous direction, whereas the bottom vessel wall remained stable. This observation indicates that the traditional vasodilation measurement in the lateral direction may result in an underestimated value.

Longitudinal imaging of vitreal hyperreflective foci in mice with acute optic nerve damage using visible-light optical coherence tomography.

Hyperreflective foci (HRFs) appear in optical coherence tomography (OCT) images of the retina and vitreous of patients with various ocular diseases. HRFs are hypothesized to be immune cells that appear in response to ischemia or tissue damage. To accurately identify HRFs and establish their clinical significance, it is necessary to replicate the detection of similar patterns in vivo in a small animal model. We combined visible-light OCT with temporal speckle averaging (TSA) to visualize and track vitreal HRFs (VHRFs) densities for three days after an optic nerve crush (ONC) injury. Resulting vis-OCT images revealed that VHRF density significantly increased approximately 10-fold at 12 h after ONC and returned to baseline three days after ONC. Additional immunohistochemistry results confirmed these VHRFs as inflammatory cells induced from optic nerve damage.

Enhancing Retinal Fundus Image Quality Assessment With Swin-Transformer-Based Learning Across Multiple Color-Spaces.

Purpose: The assessment of retinal image (RI) quality holds significant importance in both clinical trials and large datasets, because suboptimal images can potentially conceal early signs of diseases, thereby resulting in inaccurate medical diagnoses. This study aims to develop an automatic method for Retinal Image Quality Assessment (RIQA) that incorporates visual explanations, aiming to comprehensively evaluate the quality of retinal fundus images (RIs). Methods: We developed an automatic RIQA system, named Swin-MCSFNet, utilizing 28,792 RIs from the EyeQ dataset, as well as 2000 images from the EyePACS dataset and an additional 1,000 images from the OIA-ODIR dataset. After preprocessing, including cropping black regions, data augmentation, and normalization, a Swin-MCSFNet classifier based on the Swin-Transformer for multiple color-space fusion was proposed to grade the quality of RIs. The generalizability of Swin-MCSFNet was validated across multiple data centers. Additionally, for enhanced interpretability, a Score-CAM-generated heatmap was applied to provide visual explanations. Results: Experimental results reveal that the proposed Swin-MCSFNet achieves promising performance, yielding a micro-receiver operating characteristic (ROC) of 0.93 and ROC scores of 0.96, 0.81, and 0.96 for the "Good," "Usable," and "Reject" categories, respectively. These scores underscore the accuracy of RIQA based on Swin-MCSF in distinguishing among the three categories. Furthermore, heatmaps generated across different RIQA classification scores and various color spaces suggest that regions in the retinal images from multiple color spaces contribute significantly to the decision-making process of the Swin-MCSFNet classifier. Conclusions: Our study demonstrates that the proposed Swin-MCSFNet outperforms other methods in experiments conducted on multiple datasets, as evidenced by the superior performance metrics and insightful Score-CAM heatmaps. Translational Relevance: This study constructs a new retinal image quality evaluation system, which will contribute to the subsequent research of retinal images.

OCTDL: Optical Coherence Tomography Dataset for Image-Based Deep Learning Methods.

Optical coherence tomography (OCT) is a non-invasive imaging technique with extensive clinical applications in ophthalmology. OCT enables the visualization of the retinal layers, playing a vital role in the early detection and monitoring of retinal diseases. OCT uses the principle of light wave interference to create detailed images of the retinal microstructures, making it a valuable tool for diagnosing ocular conditions. This work presents an open-access OCT dataset (OCTDL) comprising over 2000 OCT images labeled according to disease group and retinal pathology. The dataset consists of OCT records of patients with Age-related Macular Degeneration (AMD), Diabetic Macular Edema (DME), Epiretinal Membrane (ERM), Retinal Artery Occlusion (RAO), Retinal Vein Occlusion (RVO), and Vitreomacular Interface Disease (VID). The images were acquired with an Optovue Avanti RTVue XR using raster scanning protocols with dynamic scan length and image resolution. Each retinal b-scan was acquired by centering on the fovea and interpreted and cataloged by an experienced retinal specialist. In this work, we applied Deep Learning classification techniques to this new open-access dataset.

Anatomical Features can Affect OCT Measures Used for Clinical Decisions and Clinical Trial Endpoints.

Purpose: To understand the association between anatomical parameters of healthy eyes and optical coherence tomography (OCT) circumpapillary retinal nerve fiber layer (cpRNFL) thickness measurements. Methods: OCT cpRNFL thickness was obtained from 396 healthy eyes in a commercial reference database (RDB). The temporal quadrant (TQ), superior quadrant (SQ), inferior quadrant (IQ), and global (G) cpRNFL thicknesses were analyzed. The commercial OCT devices code these values based on percentiles (red, <1%; yellow, ≥1% and <5%), after taking age and disc area into consideration. Four anatomical parameters were assessed: fovea-to-disc distance, an estimate of axial length, and the locations of the superior and the inferior peaks of the cpRNFL thickness curve. Pearson correlation values were obtained for the parameters and the thickness measures of each of the four cpRNFL regions, and t-tests were performed between the cpRNFL thicknesses coded as abnormal (red or yellow, <5%) versus normal (≥5%). Results: For each of the four anatomical parameters, the correlation with the thickness of one or more of the TQ, SQ, IQ, and G regions exceeded the correlation with age or disc area. All four parameters were significantly (P < 0.001) associated with the abnormal cpRNFL values. The significant parameters were not the same for the different regions; for example, a parameter could be negatively correlated for the TQ but positively correlated with the SQ or IQ. Conclusions: In addition to age and disc area, which are used for inferences in normative databases, four anatomical parameters are associated with cpRNFL thickness. Translational Relevance: Taking these additional anatomical parameters into consideration should aid diagnostic accuracy.

Thinning of the retinal nerve fiber and choroidal layers in adolescents with anorexia nervosa: A controlled study.

PURPOSE: We aimed to evaluate retinal nerve fiber and choroidal layer alterations in adolescents with anorexia nervosa using spectral-domain optical coherence tomography. METHODS: Thirty patients with anorexia nervosa and 30 healthy adolescents aged 12-18 years were included in this study. Their age, sex, body mass index, anorexia nervosa type, disease duration, and spectral-domain optical coherence tomography data were recorded. RESULTS: Central macular thickness and retinal nerve fiber layer thickness in the temporal and inferior regions were significantly lesser in patients with anorexia than in healthy controls (p<0.05). Moreover, significant choroidal thinning around the foveal and subfoveal regions in patients with anorexia was observed (p<0.05). In addition, a statistically significant relation between the increase in disease duration and the thinning of the inferior retinal nerve fiber layer was detected (p<0.05). CONCLUSION: The retinal nerve fiber layer and choroidal layer thicknesses were lesser in patients with anorexia than in healthy controls. Screening for retinal indices might prevent the development of irreversible retinal pathologies in adolescents with anorexia nervosa. In addition, thinning of the retinal nerve fiber and choroidal layers could reflect structural or functional changes in the brain of adolescents with anorexia nervosa.

Evaluation of Increase in Retinal Thickness as Diagnostic Marker in Central Artery Occlusion.

PURPOSE: To evaluate the increase in retinal thickness as a marker in predicting the onset of central retinal artery occlusions. METHODS: Retrospective clinical study conducted at one Swiss hospital. Electronic records were filtered for patients with artery occlusions. Optical coherence tomography data, including time between the imaging and ischemic event, were reviewed. Increase in relative retinal thickness was measured, defined as an increase in retinal thickness compared to the unaffected partner eye. This was correlated with the time from symptom onset. A cutoff value of relative increase of < 24.5% was applied, as suggested in previous studies. The results were compared to the time gathered from the electronic records, and sensitivity, specificity, positive predictive value as well as negative predictive value were calculated for predicting an ischemia time of < 4.5 h. RESULTS: Forty-two eyes from 41 patients with central artery occlusions were identified. Fourteen were female. Mean age was 66.4 ± 15.8 years. Initial corrected visual acuity was 2.41 ± 0.68 logMAR, and 2.13 ± 0.87 logMAR at the last follow-up (p > 0.05). Of eyes with a visual acuity of counting fingers (n = 38) or worse, 89.5% showed no improvement during follow-up, while eyes with logMAR 1 or better (n = 4) improved. Thirteen eyes (13 patients) presented within 4.5 h of the ischemic event. Four patients received i. v. thrombolysis, with visual recovery in one. In 12 eyes with an ischemia time of < 4.5 h, relative increase was below 24.5%. In the remaining 29 eyes with > 4.5 h, relative increase was below 24.5% in 4 eyes and above 24.5% in 25 eyes. This yielded a sensitivity of 92.3%, a specificity of 86.2%, with a positive predictive value of 75.0% and a negative predictive value of 96.2%. CONCLUSION: Central retinal artery occlusion is associated with severe vision loss. There is no current established therapy. Parameters that objectify the presence of a therapeutic window for thrombolysis are gaining in importance as patient history is often imprecise. Relative retinal thickness increase proved a noninvasive imaging parameter demonstrating adequate performance in detecting patients within the therapeutic window of thrombolysis. Further investigation of this parameter in central retinal occlusion is warranted.

A new intelligent system based deep learning to detect DME and AMD in OCT images.

Optical Coherence Tomography (OCT) is widely recognized as the leading modality for assessing ocular retinal diseases, playing a crucial role in diagnosing retinopathy while maintaining a non-invasive modality. The increasing volume of OCT images underscores the growing importance of automating image analysis. Age-related diabetic Macular Degeneration (AMD) and Diabetic Macular Edema (DME) are the most common cause of visual impairment. Early detection and timely intervention for diabetes-related conditions are essential for preventing optical complications and reducing the risk of blindness. This study introduces a novel Computer-Aided Diagnosis (CAD) system based on a Convolutional Neural Network (CNN) model, aiming to identify and classify OCT retinal images into AMD, DME, and Normal classes. Leveraging CNN efficiency, including feature learning and classification, various CNN, including pre-trained VGG16, VGG19, Inception_V3, a custom from scratch model, BCNN (VGG16) 2 , BCNN (VGG19) 2 , and BCNN (Inception_V3) 2 , are developed for the classification of AMD, DME, and Normal OCT images. The proposed approach has been evaluated on two datasets, including a DUKE public dataset and a Tunisian private dataset. The combination of the Inception_V3 model and the extracted feature from the proposed custom CNN achieved the highest accuracy value of 99.53% in the DUKE dataset. The obtained results on DUKE public and Tunisian datasets demonstrate the proposed approach as a significant tool for efficient and automatic retinal OCT image classification.

Hemoglobin levels are associated with retinal vascular caliber in a middle-aged birth cohort.

Vascular and neural structures of the retina can be visualized non-invasively and used to predict ocular and systemic pathologies. We set out to evaluate the association of hemoglobin (Hb) levels within the national reference interval with retinal vascular caliber, optical coherence tomography (OCT) and visual field (VF) parameters in the Northern Finland 1966 Birth Cohort (n = 2319, 42.1% male, average age 47 years). The studied parameters were evaluated in Hb quintiles and multivariable linear regression models. The lowest Hb quintile of both sexes presented the narrowest central retinal vein equivalent (CRVE) and the healthiest cardiometabolic profile compared to the other Hb quintiles. In the regression models, CRVE associated positively with Hb levels in both sexes, (Bmales = 0.068 [0.001; 0.135], Bfemales = 0.087 [0.033; 0.140]), after being adjusted for key cardiometabolic and inflammatory parameters, smoking status, and fellow vessel caliber. No statistically significant associations of Hb levels with central retinal artery equivalent, OCT or VF parameters were detected. In conclusion, Hb levels were positively and specifically associated with CRVE, indicating that Hb levels are an independent factor affecting CRVE and the effect is in parallel with established risk factors for cardiometabolic diseases.

Adaptive Optics Retinal Imaging in RDH12-Associated Early Onset Severe Retinal Dystrophy.

Purpose: RDH12 is among the most common genes found in individuals with early-onset severe retinal (EOSRD). Adaptive optics scanning light ophthalmoscopy (AOSLO) enables resolution of individual rod and cone photoreceptors in the retina. This study presents the first AOSLO imaging of individuals with RDH12-associated EOSRD. Methods: Case series of patients who attended Moorfields Eye Hospital (London, UK). Spectral-domain optical coherence tomography, near-infrared reflectance (NIR), and blue autofluorescence imaging were analyzed. En face image sequences of photoreceptors were recorded using either of two AOSLO modalities. Cross-sectional analysis was undertaken for seven patients and longitudinal analysis for one patient. Results: Nine eyes from eight patients are presented in this case series. The mean age at the time of the assessment was 11.2 ± 6.5 years of age (range 7-29). A subfoveal continuous ellipsoid zone (EZ) line was present in eight eyes. Posterior pole AOSLO revealed patches of cone mosaics. Average cone densities at regions of interest 0.5° to the fovea ranged from 12,620 to 23,660 cells/mm2, whereas intercell spacing ranged from 7.0 to 9.7 µm. Conclusions: This study demonstrates that AOSLO can provide useful high-quality images in patients with EOSRD, even during childhood, with nystagmus, and early macular atrophy. Cones at the posterior pole can appear as scattered islands or, possibly later in life, as a single subfoveal conglomerate. Detailed image analysis suggests that retinal pigment epithelial stress and dysfunction may be the initial step toward degeneration, with NIR being a useful tool to assess retinal well-being in RDH12-associated EOSRD.

The clinical evaluation of a widefield lens to expand the field of view in optical coherence tomography (OCT-A).

This study aimed to evaluate the clinical benefits of incorporating a widefield lens (WFL) in optical coherence tomography angiography (OCT-A) in patients with retinal vascular diseases in comparison to standard single-shot OCT-A scans. Sixty patients with retinal vascular diseases including diabetic retinopathy (DR) and retinal vein occlusion (RVO) were recruited. OCT-A imaging (PlexElite 9000) with and without WFL was performed in randomized order. The assessment included patient comfort, time, field of view (FoV), image quality and pathology detection. Statistical analysis included paired t-tests, Mann-Whitney U-tests and Bonferroni correction for multiple tests, with inter-grader agreement using the kappa coefficient. Using a WFL did not lead to statistically significant differences in DR and RVO group test times. Patient comfort remained high, with similar responses for WFL and non-WFL measurements. The WFL notably expanded the scan field (1.6× FoV increase), enhancing peripheral retinal visibility. However, image quality varied due to pathology and eye dominance, affecting the detection of peripheral issues in RVO and DR cases. The use of a WFL widens the scan field, aiding vascular retinal disease imaging with minor effects on comfort, time, and image quality. Further enhancements are needed for broader view angles, enabling improved quantification of non-perfused areas and more reliable peripheral proliferation detection.

Novel optical coherence tomography findings in a patient with methanol toxicity: a case report.

We report a case of acute methanol toxicity with unique optical coherence tomography findings. A 56-year-old man was referred to our ophthalmology clinic with a history of handmade vodka consumption and vision loss. On ophthalmologic examination, his vision was 20/100 in his right eye and 20/200 in his left eye. Bilateral mild optic disk hyperemia was detected on fundus examination. Because of the severity of systemic symptoms in such cases, it is very difficult to include optical coherence tomography in the ophthalmologic examination. However, we managed to perform optical coherence tomography and recorded shallow subretinal fluid and a prominent middle limiting membrane sign as acute retinal structural changes in the patient. The patient was treated with hemodialysis, intravenous ethanol, and sodium bicarbonate. On the fourth day of treatment, visual acuity improved to 20/20 in both eyes. In addition, the prominent middle limiting membrane sign and subretinal fluid disappeared. In this unusual case, retinal pigment epithelium damage and retinal ischemia may have contributed to the prominent middle limiting membrane and subretinal fluid, which are novel optical coherence tomography findings of methanol toxicity.

Comparative In Vivo Imaging of Retinal Structures in Tree Shrews, Humans, and Mice.

Rodent models, such as mice and rats, are commonly used to examine retinal ganglion cell damage in eye diseases. However, as nocturnal animals, rodent retinal structures differ from primates, imposing significant limitations in studying retinal pathology. Tree shrews (Tupaia belangeri) are small, diurnal paraprimates that exhibit superior visual acuity and color vision compared with mice. Like humans, tree shrews have a dense retinal nerve fiber layer (RNFL) and a thick ganglion cell layer (GCL), making them a valuable model for investigating optic neuropathies. In this study, we applied high-resolution visible-light optical coherence tomography to characterize the tree shrew retinal structure in vivo and compare it with that of humans and mice. We quantitatively characterize the tree shrew's retinal layer structure in vivo, specifically examining the sublayer structures within the inner plexiform layer (IPL) for the first time. Next, we conducted a comparative analysis of retinal layer structures among tree shrews, mice, and humans. We then validated our in vivo findings in the tree shrew inner retina using ex vivo confocal microscopy. The in vivo and ex vivo analyses of the shrew retina build the foundation for future work to accurately track and quantify the retinal structural changes in the IPL, GCL, and RNFL during the development and progression of human optic diseases.

Morphological differences of the neuroretinal rim between temporally tilted and non-tilted optic discs in healthy eyes.

This study aimed to compare morphological differences of the neuroretinal rim between the temporally tilted and non-tilted optic discs in healthy eyes. We prospectively enrolled participants aged 20-40 years with temporally tilted or non-tilted optic discs. The optic nerve head parameters were analyzed using spectral domain-optical coherence tomography. The angle between the Bruch's membrane opening (BMO) plane and BMO-minimum rim width (BMO-MRW) was termed "BMO-MRW angle". Peripapillary retinal nerve fiber layer thickness (pRNFLT) and BMO-based parameters were compared between the temporally tilted and non-tilted disc groups. As a result, 55 temporally tilted disc eyes and 38 non-tilted disc eyes were analyzed. Global pRNFLT, global BMO-MRW, and total BMO-minimum rim area (BMO-MRA) were similar between the two groups (p = 0.138, 0.161, and p = 0.410, respectively). In the sectoral analysis, temporally tilted disc group exhibited thicker BMO-MRW in the temporal sector (p = 0.032) and thinner in the nasal superior and nasal sectors (p = 0.025 and p = 0.002, respectively). Temporally tilted disc group showed larger BMO-MRA in the temporal, temporal superior, and temporal inferior sectors (p < 0.001, p < 0.001, and p < 0.016, respectively), alongside a higher BMO-MRW angle in the temporal sector and lower in the nasal superior and nasal sectors. In conclusion, the neuroretinal rim, represented by BMO-MRW and BMO-MRA, showed morphological differences between temporally tilted and non-tilted optic discs in healthy eyes. BMO-MRW and BMO-MRA showed temporalization in the same manner as pRNFLT in the temporally tilted disc eyes. The BMO-MRW angle showed that in temporally tilted disc eyes, optic nerve fibers met the BMO plane steeply in the nasal sector and gently in the temporal sector than in non-tilted disc eyes, suggesting potential stress region of optic nerve fibers in temporally tilted disc eyes.

The Impact of Photopigment Bleaching on the Human Rod Photoreceptor Subretinal Space Measured Via Optical Coherence Tomography.

Purpose: The purpose of this study was to investigate rod photopigment bleaching-driven intrinsic optical signals (IOS) in the human outer retina and its measurement repeatability based on a commercial optical coherence tomography (OCT) platform. Methods: The optical path length of the rod photoreceptor subretinal space (SRS), that is, the distance between signal bands of rod outer segment tips and retinal pigment epithelium, was measured in 15 healthy subjects in ambient light and during a long-duration bleaching white-light exposure. Results: On 2 identical study days (day 1 and day 2 [D1 and D2]), light stimulation resulted in a significant decrease in rod SRS by 21.3 ± 7.6% and 19.8 ± 8.5% (both P < 0.001), respectively. The test-retest reliability of the SRS maximum change of an individual subject was moderate for single measures (intraclass correlation coefficient [ICC] = 0.730, 95% confidence interval [CI] = 0.376, 0.900, P < 0.001) and good for average measures (ICC = 0.844, 95% CI = 0.546, 0.947, P < 0.001). The mean area under the stimulus response curve with values of 14.8 ± 9.4 and 15.5 ± 7.5 µm × minutes (P = 0.782) showed excellent agreement between the stimulus response on D1 and D2. Intermittent dark adaptation of the retina led to an initial increase of the SRS by 6.1% (P = 0.018) and thereafter showed a decrease toward baseline, despite continued dark adaptation. Conclusions: The data indicate the potential of commercial OCT in measuring slow IOS in the outer retina suggesting that the rod SRS could serve as a biomarker for photoreceptor function. The presented approach could provide an easily implementable clinical tool for the early detection of diseases affecting photoreceptor health.

PGKD-Net: Prior-guided and Knowledge Diffusive Network for Choroid Segmentation.

The thickness of the choroid is considered to be an important indicator of clinical diagnosis. Therefore, accurate choroid segmentation in retinal OCT images is crucial for monitoring various ophthalmic diseases. However, this is still challenging due to the blurry boundaries and interference from other lesions. To address these issues, we propose a novel prior-guided and knowledge diffusive network (PGKD-Net) to fully utilize retinal structural information to highlight choroidal region features and boost segmentation performance. Specifically, it is composed of two parts: a Prior-mask Guided Network (PG-Net) for coarse segmentation and a Knowledge Diffusive Network (KD-Net) for fine segmentation. In addition, we design two novel feature enhancement modules, Multi-Scale Context Aggregation (MSCA) and Multi-Level Feature Fusion (MLFF). The MSCA module captures the long-distance dependencies between features from different receptive fields and improves the model's ability to learn global context. The MLFF module integrates the cascaded context knowledge learned from PG-Net to benefit fine-level segmentation. Comprehensive experiments are conducted to evaluate the performance of the proposed PGKD-Net. Experimental results show that our proposed method achieves superior segmentation accuracy over other state-of-the-art methods. Our code is made up publicly available at: https://github.com/yzh-hdu/choroid-segmentation.