Design and Implementation of AI based Ophthalmological Diagnosis of Eye Disease using Optical Coherance Tomography OCT machine and Image Processing
Early detection and appropriate treatment of eye diseases are of great significance to prevent vision loss and promote living quality. Conventional diagnostic methods depend upon physicians? professional experience and knowledge, which lead to high misdiagnosis rate and huge waste of medical data. D
2025-06-28 16:31:42 - Adil Khan
Design and Implementation of AI based Ophthalmological Diagnosis of Eye Disease using Optical Coherance Tomography OCT machine and Image Processing
Project Area of Specialization Artificial IntelligenceProject SummaryEarly detection and appropriate treatment of eye diseases are of great significance to prevent vision loss and promote living quality. Conventional diagnostic methods depend upon physicians’ professional experience and knowledge, which lead to high misdiagnosis rate and huge waste of medical data. Deep integration of ophthalmology and artificial intelligence (AI) has the potential to revolutionize current disease diagnostic pattern and generate a significant clinical impact. In this project, the detailed report will be generated by the development and validation of a fully data-driven artificial intelligence–based Convolutional Neural Network (CNN) algorithm that can be used to screen Spectral Domain Optical Coherence Tomography (SD-OCT) photographs obtained from diabetic patients to automatically identify and detect macula fluid i.e. Intra-retinal Cystoid (IRC) fluid (see Fig. 1) in the most occurring eye disease like Diabetic Macula Edema (DME) with high reliability. Moreover, this project will generate an automatic report which will assist clinicians in monitoring the progression of IRC fluid. This would aid in reducing the rate of vision loss, enabling timely and accurate diagnoses. The implementation of such an algorithm could reduce drastically the rate of vision loss attributed to DME. The block diagram of project is show in Fig. 2.
Fig. 1: Cross sectional area of Macula filled with Fluid and Normal Eye image obtained from OCT
Fig. 2: Conception diagram of the project
Project Objectives- To design a deep learning and image processing based application to process OCT images and implement best possible algorithm to predict IRC fluid regions in DME eye disease.
- To automatically generate report which will show a diagnosed percentage of disease as an accurately diagnosed result compared to the conventional approach used by physicians.
- To validate the results by expert readers trained for retinal analyses between automated and manual delineation of fluid features.
- To implement proposed project on standalone device.
This project will be implemented through detection and identification of the presence of IRC fluid regions of DME disease for each location (pixel) in the OCT image with the aid of developing semantic segmentation. In our case the goal of semantic image segmentation is to label each pixelof an image with a corresponding class of what is being represented i.e. IRC fluid. A method is adopted based on convolutional neural networks. We apply deep learning, a state-of-the-art machine learning technique that learns the mapping from OCT images to pixel-level class label based on large amounts of labelled training data. Deep learning models allow one to learn meaningful abstract data representations. Following the semantic segmentation approach, the neural network maps an input image of a specific size to an image of corresponding class labels of the same size. The evaluation of performance and the pixel-level segmentation accuracy perform on the basis of pixel-wise IRC fluid segmentations by the software and corresponding ground truth annotations by reading experts.
The proposed neural network comprises 2 processing components, an encoder/contraction that transforms an input image into an abstract representation and a decoder/expansion that maps the abstract representation to an image of clinical single class label assigning each pixel a class i.e. IRC fluid. The mapping of the encoder from raw images to abstract representations (embeddings) is not computed on the basis of pre-specified mathematic descriptions (handcrafted features), but the encoder parameters will automatically learn solely on the basis of annotated data used during training. The data embedding learned is optimize in such a way that will best for the generation of a corresponding image of class labels. The mapping of the encoder from raw images to the data embedding needed to generate the label image, and the mapping of the decoder from the embedding to a full input resolution label image will learn simultaneously (end-to-end). The encoder and the decoder comprised a set of computing blocks (layers), where the layers of the decoder virtually inversed the operations of the encoder conditioned by the low-dimensional embedding learned by the encoder. Its architecture is modified and extended to work with fewer training images of any size and to yield more precise segmentations.
Fig. 3: Illustration of Data (OCT images) flow internally through the Proposed Model.
Benefits of the ProjectThe presence of macular fluid represents the most important diagnostic retreatment criterion in the management of patients with exudative macular disease, and the evaluation of the fluid status on OCT has become a routine task not only for retina specialists but also for ophthalmologists in practice globally.
- Our method uses state-of-the-art artificial intelligence technology to solve the practical task of active fluid detection and quantification.
- On the basis of standard imaging, a reliable evaluation of leakage activity is often controversial, which matters particularly if retreatment is aimed at stabilization of disease activity such as in DME. A manual inspection of large OCT datasets in busy clinics is inherently impractical and prone to errors. Our automated method presents an attractive tool to monitor OCT datasets and direct the clinician’s attention to those images requiring detailed analysis.
- It will also be helpful in order to predict the early stage symptoms of DME depending on the area occupied by fluid, which if controlled can help in stopping that disease from further expansion.
- Data driven analysis and subsequent optimization for the number of other eye related diseases.
- A home-grown e-health solution which will enable the OCT technician to readily generate standardized diagnostic report for the patients of DME.
- It aims to have potential impact on the local industry and economy helping to reduce the human error produced by the Physicians that depend on their knowledge and personal experience
- By commercializing, it will help doctors as well as patients to deal with eye check-up more accurately.
- To help medical students to understand eye disease occurring in eye.
Our Final Product will consist of a real-time stand-alone device (Raspberry Pi / ARM Microprocessor), that will take input images through the OCT Machine and process them through image processing A.I. algorithm and then Deep Learning Model will generate a report accordingly.
- A detailed report will be generated, showing different eye parameters (eye layers, detected fluid, area occupied by fluid inside the macula, severity level etc.)
- The report generated by the project will be showing the final results, i.e. the eyes are affected or not, if yes, the presence of macula fluid will confirm the disease.
- In the report, the amount of macula fluid will also be shown, which is earlier measured by A.I. Model, this feature will elaborate how much severe, the disease is.
- Evaluation of retinal thickening caused by fluid accumulation, in DME are carried out in the final report.
- Each patient’s data will be saved and that data will be exported to eye specialist for further treatment.
The overview of report layout is shown in Fig. 4
Fig. 4: Auto Generated Report through the Proposed Model.
Final Deliverable of the Project HW/SW integrated systemCore Industry HealthOther Industries IT , Medical , Health Core Technology Artificial Intelligence(AI)Other Technologies Artificial Intelligence(AI)Sustainable Development Goals Good Health and Well-Being for PeopleRequired Resources
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 75097 | |||
| GeForce GTX 1650 AMP! Edition | Equipment | 1 | 27399 | 27399 |
| Raspberry pi 4 | Equipment | 2 | 10000 | 20000 |
| Kingston MicroSD 8GB | Equipment | 2 | 599 | 1198 |
| HDMI to HDMI cable | Equipment | 2 | 180 | 360 |
| Raspberry pi 4 Power Supply | Equipment | 2 | 300 | 600 |
| Raspberry 4 casing | Equipment | 2 | 1020 | 2040 |
| STM32MP157A | Equipment | 1 | 13500 | 13500 |
| Consultation fee of eye specialist (for testing) | Miscellaneous | 5 | 1000 | 5000 |
| Travelling for surveys and consultations | Miscellaneous | 2 | 2500 | 5000 |