Robust Disease Diagnosis System using Fundoscopic Retinal Vessel Segmentation
The ability to accurately segment retinal blood vessels in the fundus is critical to the diagnosis of fundus disorders. We can choose the retinal vessel segmentation to improve the thick and thin vessels of the retina, and this method help in the diagnosis. We're utilizing the U-Net because traditio
2025-06-28 16:28:59 - Adil Khan
Robust Disease Diagnosis System using Fundoscopic Retinal Vessel Segmentation
Project Area of Specialization Artificial IntelligenceProject SummaryThe ability to accurately segment retinal blood vessels in the fundus is critical to the diagnosis of fundus disorders. We can choose the retinal vessel segmentation to improve the thick and thin vessels of the retina, and this method help in the diagnosis. We're utilizing the U-Net because traditional retinal segmentation has the incomplete resection of micro-vessels and results in significant mis-segmentation. In addition to reducing the network's complexity, the proposed technique enhanced segmentation accuracy. The addition of residual elements will contribute to the network's overall depth and performance improvement. Using the DRIVE dataset, the proposed technique was more advantageous than existing methods. This algorithm will be used to a wider range of medical specialties in the future. And then, we may discover out numerous illnesses based on the supplied dataset in fundus photographs. We can design a front end that is based on detection. We may enter the input and obtain the results in the prediction and segmentation of those photographs and figure out the ailment. We can improve the prior segmentation and detect several illnesses on different datasets, and it is the right model that sends to the doctor.
Project ObjectivesThe main goal of this dissertation is to study and analyze different approaches based on deep learning techniques for the segmentation of retinal blood vessels. In order to do so, different designs and architectures of CNNs will be studied and analyzed, as their results and performance are evaluated and compared with the available algorithms. One other important objective of this work is to study and evaluate the different techniques that have been used for vessel segmentation, based on machine learning, and how these can be combined with the deep learning approaches: by analyzing the features that the learned models are using to perform classification and combining them with different machine learning techniques, another objective is to propose a solution or set of solutions to perform the retinal vessel segmentation, using this methodology of work.
Project Implementation MethodFirst of all, we can collect the two types of datasets the Digital Retinal Images for Vessel Extraction (DRIVE) and HRF (High-Resolution Fundus) that are used for the vessels segmentation Now we can do the classification and then go to the segmentation phase.
After segmentation, we can collect the dataset name is the Retinal fundus multi-Disease image Dataset (RFMID) to find out the disease. we can use this for classification and then we can be trained, test, and validation of the model. we can define multiple diseases and then predict the disease.
Benefits of the ProjectWhen it comes to retinal vascular segmentation algorithms, most of them are focused on automated detection linked to retinopathy, which is widely recognized to be the primary cause of blindness in the modern era. If diabetic retinopathy is discovered in its initial stages, therapies may be administered to avoid loss of vision associated with the illness. Because of this, several writers have suggested a variety of distinct blood artery segmentation methodologies that are based on various techniques. The algorithm's complexity and segmentation are distinct in terms of their structure. A variety of vascular segmentation techniques are investigated and applied in this work, and their results are evaluated in relation to the obtained findings.
Technical Details of Final DeliverableA platform that is used for the project is
- Google Colab
- Jupyter Notebook
- Kaggle
The algorithm that is used for the project is
- CNN
- U-Net
- VGG16
| Item Name | Type | No. of Units | Per Unit Cost (in Rs) | Total (in Rs) |
|---|---|---|---|---|
| Total in (Rs) | 15480 | |||
| AMD RX 580 | Equipment | 0 | 63000 | 0 |
| Udemy Course | Equipment | 1 | 15480 | 15480 |