Machine-Identified Patterns of Visual Field Loss and an Association with Rapid Progression in the Ocular Hypertension Treatment Study

Large language models as assistance for glaucoma surgical cases: a ChatGPT vs. Google Gemini comparison

PURPOSE

The aim of this study was to define the capability of ChatGPT-4 and Google Gemini in analyzing detailed glaucoma case descriptions and suggesting an accurate surgical plan.

METHODS

Retrospective analysis of 60 medical records of surgical glaucoma was divided into "ordinary" (n = 40) and "challenging" (n = 20) scenarios. Case descriptions were entered into ChatGPT and Bard's interfaces with the question "What kind of surgery would you perform?" and repeated three times to analyze the answers' consistency. After collecting the answers, we assessed the level of agreement with the unified opinion of three glaucoma surgeons. Moreover, we graded the quality of the responses with scores from 1 (poor quality) to 5 (excellent quality), according to the Global Quality Score (GQS) and compared the results.

RESULTS

ChatGPT surgical choice was consistent with those of glaucoma specialists in 35/60 cases (58%), compared to 19/60 (32%) of Gemini (p = 0.0001). Gemini was not able to complete the task in 16 cases (27%). Trabeculectomy was the most frequent choice for both chatbots (53% and 50% for ChatGPT and Gemini, respectively). In "challenging" cases, ChatGPT agreed with specialists in 9/20 choices (45%), outperforming Google Gemini performances (4/20, 20%). Overall, GQS scores were 3.5 ± 1.2 and 2.1 ± 1.5 for ChatGPT and Gemini (p = 0.002). This difference was even more marked if focusing only on "challenging" cases (1.5 ± 1.4 vs. 3.0 ± 1.5, p = 0.001).

CONCLUSION

ChatGPT-4 showed a good analysis performance for glaucoma surgical cases, either ordinary or challenging. On the other side, Google Gemini showed strong limitations in this setting, presenting high rates of unprecise or missed answers.

Artificial intelligence for glaucoma: state of the art and future perspectives

PURPOSE OF REVIEW

To address the current role of artificial intelligence (AI) in the field of glaucoma.

RECENT FINDINGS

Current deep learning (DL) models concerning glaucoma diagnosis have shown consistently improving diagnostic capabilities, primarily based on color fundus photography and optical coherence tomography, but also with multimodal strategies. Recent models have also suggested that AI may be helpful in detecting and estimating visual field progression from different input data. Moreover, with the emergence of newer DL architectures and synthetic data, challenges such as model generalizability and explainability have begun to be tackled.

SUMMARY

While some challenges remain before AI is routinely employed in clinical practice, new research has expanded the range in which it can be used in the context of glaucoma management and underlined the relevance of this research avenue.

Lessons From The Glaucoma Foundation Think Tank 2023: A Patient-Centric Approach to Glaucoma

PRCIS

The main takeaways also included that BIG DATA repositories and AI are important combinatory tools to foster novel strategies to prevent and stabilize glaucoma and, in the future, recover vision loss from the disease.

PURPOSE

To summarize the main topics discussed during the 28th Annual Glaucoma Foundation Think Tank Meeting "A Patient-Centric Approach to Glaucoma" held in New York on June 9 and 10, 2023.

METHODS

The highlights of the sessions on BIG DATA, genetics, modifiable lifestyle risk factors, female sex hormones, and neuroprotection in the field of primary open angle glaucoma (POAG) were summarized.

RESULTS

The researchers discussed the importance of BIG DATA repositories available at national and international levels for POAG research, including the United Kingdom Biobank. Combining genotyped large cohorts worldwide, facilitated by artificial intelligence (AI) and machine-learning approaches, led to the milestone discovery of 312 genome-wide significant disease loci for POAG. While these loci could be combined into a polygenic risk score with clinical utility, Think Tank meeting participants also provided analytical epidemiological evidence that behavioral risk factors modify POAG polygenetic risk, citing specific examples related to caffeine and alcohol use. The impact of female sex hormones on POAG pathophysiology was discussed, as was neuroprotection and the potential use of AI to help mitigate specific challenges faced in clinical trials and speed approval of neuroprotective agents.

CONCLUSIONS

The experts agreed on the importance of genetics in defining individual POAG risk and highlighted the additional crucial role of lifestyle, gender, blood pressure, and vascular risk factors. The main takeaways also included that BIG DATA repositories and AI are important combinatory tools to foster novel strategies to prevent and stabilize glaucoma and, in the future, recover vision loss from the disease.

Machine Learning-Derived Baseline Visual Field Patterns Predict Future Glaucoma Onset in the Ocular Hypertension Treatment Study

Purpose

The Ocular Hypertension Treatment Study (OHTS) identified risk factors for primary open-angle glaucoma (POAG) in patients with ocular hypertension, including pattern standard deviation (PSD). Archetypal analysis, an unsupervised machine learning method, may offer a more interpretable approach to risk stratification by identifying patterns in baseline visual fields (VFs).

Methods

There were 3272 eyes available in the OHTS. Archetypal analysis was applied using 24-2 baseline VFs, and model selection was performed with cross-validation. Decomposition coefficients for archetypes (ATs) were calculated. A penalized Cox proportional hazards model was implemented to select discriminative ATs. The AT model was compared to the OHTS model. Associations were identified between ATs with both POAG onset and VF progression, defined by mean deviation change per year.

Results

We selected 8494 baseline VFs. Optimal AT count was 19. The highest prevalence ATs were AT9, AT11, and AT7. The AT-based prediction model had a C-index of 0.75 for POAG onset. Multivariable models demonstrated that a one-interquartile range increase in the AT5 (hazard ratio [HR] = 1.14; 95% confidence interval [CI], 1.04-1.25), AT8 (HR = 1.22; 95% CI, 1.09-1.37), AT15 (HR = 1.26; 95% CI, 1.12-1.41), and AT17 (HR = 1.17; 95% CI, 1.03-1.31) coefficients conferred increased risk of POAG onset. AT5, AT10, and AT14 were significantly associated with rapid VF progression. In a subgroup analysis by high-risk ATs (>95th percentile or <75th percentile coefficients), PSD lost significance as a predictor of POAG in the low-risk group.

Conclusions

Baseline VFs, prior to detectable glaucomatous damage, contain occult patterns representing early changes that may increase the risk of POAG onset and VF progression in patients with ocular hypertension. The relationship between PSD and POAG is modified by the presence of high-risk patterns at baseline. An AT-based prediction model for POAG may provide more interpretable glaucoma-specific information in a clinical setting.

Artificial intelligence in glaucoma: opportunities, challenges, and future directions

Artificial intelligence (AI) has shown excellent diagnostic performance in detecting various complex problems related to many areas of healthcare including ophthalmology. AI diagnostic systems developed from fundus images have become state-of-the-art tools in diagnosing retinal conditions and glaucoma as well as other ocular diseases. However, designing and implementing AI models using large imaging data is challenging. In this study, we review different machine learning (ML) and deep learning (DL) techniques applied to multiple modalities of retinal data, such as fundus images and visual fields for glaucoma detection, progression assessment, staging and so on. We summarize findings and provide several taxonomies to help the reader understand the evolution of conventional and emerging AI models in glaucoma. We discuss opportunities and challenges facing AI application in glaucoma and highlight some key themes from the existing literature that may help to explore future studies. Our goal in this systematic review is to help readers and researchers to understand critical aspects of AI related to glaucoma as well as determine the necessary steps and requirements for the successful development of AI models in glaucoma.

The Use of ChatGPT to Assist in Diagnosing Glaucoma Based on Clinical Case Reports

INTRODUCTION

The purpose of this study was to evaluate the capabilities of large language models such as Chat Generative Pretrained Transformer (ChatGPT) to diagnose glaucoma based on specific clinical case descriptions with comparison to the performance of senior ophthalmology resident trainees.

METHODS

We selected 11 cases with primary and secondary glaucoma from a publicly accessible online database of case reports. A total of four cases had primary glaucoma including open-angle, juvenile, normal-tension, and angle-closure glaucoma, while seven cases had secondary glaucoma including pseudo-exfoliation, pigment dispersion glaucoma, glaucomatocyclitic crisis, aphakic, neovascular, aqueous misdirection, and inflammatory glaucoma. We input the text of each case detail into ChatGPT and asked for provisional and differential diagnoses. We then presented the details of 11 cases to three senior ophthalmology residents and recorded their provisional and differential diagnoses. We finally evaluated the responses based on the correct diagnoses and evaluated agreements.

RESULTS

The provisional diagnosis based on ChatGPT was correct in eight out of 11 (72.7%) cases and three ophthalmology residents were correct in six (54.5%), eight (72.7%), and eight (72.7%) cases, respectively. The agreement between ChatGPT and the first, second, and third ophthalmology residents were 9, 7, and 7, respectively.

CONCLUSIONS

The accuracy of ChatGPT in diagnosing patients with primary and secondary glaucoma, using specific case examples, was similar or better than senior ophthalmology residents. With further development, ChatGPT may have the potential to be used in clinical care settings, such as primary care offices, for triaging and in eye care clinical practices to provide objective and quick diagnoses of patients with glaucoma.

Multi-Dataset Comparison of Vision Transformers and Convolutional Neural Networks for Detecting Glaucomatous Optic Neuropathy from Fundus Photographs

Glaucomatous optic neuropathy (GON) can be diagnosed and monitored using fundus photography, a widely available and low-cost approach already adopted for automated screening of ophthalmic diseases such as diabetic retinopathy. Despite this, the lack of validated early screening approaches remains a major obstacle in the prevention of glaucoma-related blindness. Deep learning models have gained significant interest as potential solutions, as these models offer objective and high-throughput methods for processing image-based medical data. While convolutional neural networks (CNN) have been widely utilized for these purposes, more recent advances in the application of Transformer architectures have led to new models, including Vision Transformer (ViT,) that have shown promise in many domains of image analysis. However, previous comparisons of these two architectures have not sufficiently compared models side-by-side with more than a single dataset, making it unclear which model is more generalizable or performs better in different clinical contexts. Our purpose is to investigate comparable ViT and CNN models tasked with GON detection from fundus photos and highlight their respective strengths and weaknesses. We train CNN and ViT models on six unrelated, publicly available databases and compare their performance using well-established statistics including AUC, sensitivity, and specificity. Our results indicate that ViT models often show superior performance when compared with a similarly trained CNN model, particularly when non-glaucomatous images are over-represented in a given dataset. We discuss the clinical implications of these findings and suggest that ViT can further the development of accurate and scalable GON detection for this leading cause of irreversible blindness worldwide.

Analysis of volumetric 3D reconstruction of lamina cribrosa images from swept-source optical coherence tomography in glaucomatous and healthy subjects

This study demonstrates the 3D visualization of the lamina cribrosa (LC) structure and its correlation with volumetric data, pore volume, and disc area in glaucomatous and non-glaucomatous eyes. The participant cohort included 65 glaucomatous and 58 non-glaucomatous eyes (13 suspected glaucoma and 45 normal). An ophthalmologist diagnosed glaucoma patients and all subjects were over 18 years old, passed a visual field test, and underwent optical coherence tomography (OCT) examinations. LC images were obtained using the DRI OCT Triton, while optic disc images were obtained from the enface image of the Cirrus HD-OCT 5000. Since LC images alone did not provide clear edge information, we used optic disc images as a reference for edge detection. To achieve this, we employed a fine-tuned model, specifically a pre-trained U-shaped Encoder-Decoder Network with Attention. This model was used to obtain a segmented mask, which was then aligned and utilized to locate the edge of the LC in the LC images. A blood vessel mask was created to remove blood vessels, as they can interfere with the accurate visualization and analysis of LC characteristics. This step allowed for the 3D reconstruction of the LC structure without the presence of blood vessels. Correlations between LC volume, pore volume, and pore volume to LC volume were calculated separately for glaucomatous and non-glaucomatous eyes. We divided the areas for considering the LC structure into three types: overall, quadrants, and 12-clock-hour sectors. Based on the experimental results, we found that the pore volume and pore-to-LC volume were different between glaucoma and normal across all areas considered. In conclusion, this research generated 3D images of the LC from OCT images using computer techniques, showcasing a microstructure that closely resembles the actual LC. Statistical methods were employed to calculate and analyze the differences observed between the two groups of samples.

Use of artificial intelligence in forecasting glaucoma progression

Artificial intelligence (AI) has been widely used in ophthalmology for disease detection and monitoring progression. For glaucoma research, AI has been used to understand progression patterns and forecast disease trajectory based on analysis of clinical and imaging data. Techniques such as machine learning, natural language processing, and deep learning have been employed for this purpose. The results from studies using AI for forecasting glaucoma progression however vary considerably due to dataset constraints, lack of a standard progression definition and differences in methodology and approach. While glaucoma detection and screening have been the focus of most research that has been published in the last few years, in this narrative review we focus on studies that specifically address glaucoma progression. We also summarize the current evidence, highlight studies that have translational potential, and provide suggestions on how future research that addresses glaucoma progression can be improved.

Assessment of visual field progression in glaucoma

PURPOSE OF REVIEW

Perimetry plays an important role in the diagnosis and management of glaucoma. This article discusses the assessment of visual field progression in patients with glaucoma.

RECENT FINDINGS

Selecting the best visual field test strategy and establishing a baseline of visual fields will assist clinicians in the detection of glaucomatous progression. Repeat testing serves to confirm or refute changes on visual field testing. More frequent testing after initial diagnosis is recommended to establish a baseline and to identify patients with rapid progression who may need more aggressive management. Statistically significant changes on event analysis can prompt examination of a patient's trend analysis to determine whether clinically significant deterioration may be occurring. Future applications of machine learning can complement existing methods of visual field interpretation.

SUMMARY

Many treated patients with glaucoma will experience visual field progression. Optimal utilization of visual field testing strategy and analytical software can help clinicians identify patients with glaucomatous progression likely to cause functional visual disability.

Follow-up studies of the classical landmark studies in Glaucoma

PURPOSE OF REVIEW

It was not until the nineteen nineties that there was scientific evidence for the primary treatment concept of glaucoma, lowering intraocular pressure. The treatment concept of lowering intraocular pressure is now proven by randomized controlled clinical trials (landmark studies). This review provides an overview of the follow-up studies to these landmark studies from the last 18 months.

RECENT FINDINGS

The 20-year data of the ocular hypertension treatment study showed a 49% incidence of primary open-angle glaucoma. The data of this landmark study was used for developing different machine learning algorithms. Bilateral disease, disc hemorrhages, and higher intraocular pressure (IOP) were risk factors for visual field deterioration in the United Kingdom Glaucoma Treatment Study. The Advanced Glaucoma Intervention Trial and the Collaborative Initial Glaucoma Treatment Study identified demographic factors, comorbidity, and clinical variables associated with visual field damage. Analysis was performed on Chinese subsets from the Laser in Glaucoma and Ocular Hypertension Study (LiGHT). Looking at all the follow-up studies to the tube shunt landmark studies, preoperative IOP, neovascular glaucoma, Ahmed implantation, and younger age were predictors of failure. A follow-up study to the Effectiveness in Angle-Closure Glaucoma in Lens Extraction Study showed that patients were ten times more likely to maintain a drop-free target IOP after lens extraction than after laser iridotomy. A superior location for iridotomy results in significantly more significant angle widening than temporal or nasal locations, as shown by a follow-up study to the Zhongshan Angle Closure Prevention Trial using OCT and gonioscopy.

SUMMARY

The number of published follow-up studies to the landmark studies in glaucoma show the ongoing development of clinical questions in management of glaucoma.

Applications of Artificial Intelligence and Deep Learning in Glaucoma

Diagnosis and detection of progression of glaucoma remains challenging. Artificial intelligence-based tools have the potential to improve and standardize the assessment of glaucoma but development of these algorithms is difficult given the multimodal and variable nature of the diagnosis. Currently, most algorithms are focused on a single imaging modality, specifically screening and diagnosis based on fundus photos or optical coherence tomography images. Use of anterior segment optical coherence tomography and goniophotographs is limited. The majority of algorithms designed for disease progression prediction are based on visual fields. No studies in our literature search assessed the use of artificial intelligence for treatment response prediction and no studies conducted prospective testing of their algorithms. Additional challenges to the development of artificial intelligence-based tools include scarcity of data and a lack of consensus in diagnostic criteria. Although research in the use of artificial intelligence for glaucoma is promising, additional work is needed to develop clinically usable tools.

Clinical Applications of Artificial Intelligence in Glaucoma

Ophthalmology is one of the major imaging-intensive fields of medicine and thus has potential for extensive applications of artificial intelligence (AI) to advance diagnosis, drug efficacy, and other treatment-related aspects of ocular disease. AI has made impressive progress in ophthalmology within the past few years and two autonomous AI-enabled systems have received US regulatory approvals for autonomously screening for mid-level or advanced diabetic retinopathy and macular edema. While no autonomous AI-enabled system for glaucoma screening has yet received US regulatory approval, numerous assistive AI-enabled software tools are already employed in commercialized instruments for quantifying retinal images and visual fields to augment glaucoma research and clinical practice. In this literature review (non-systematic), we provide an overview of AI applications in glaucoma, and highlight some limitations and considerations for AI integration and adoption into clinical practice.