International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
Convolutional Neural Network for Advanced Glaucoma Detection Dinesh Sai Kumar Pilla1, Eesha Smitha Ravella2, Kolli Naga Sai Venkata Rohit3, Udheep Perla4, Jaswanth Kolli5, Dr Rama Narasinga Rao Manda6 1,2,3,4,5Student, GITAM (Deemed to be University), Visakhapatnam, Andhra Pradesh, India
6Professor, Dept. of CSE, GITAM (Deemed to be University), Visakhapatnam, Andhra Pradesh, India
---------------------------------------------------------------------***---------------------------------------------------------------------
Abstract – Glaucoma, a leading cause of irreversible
has breathed new hope into the realm of automated disease detection, particularly in the analysis of retinal images, where glaucoma often leaves its subtle traces. This undertaking embarks on a quest to explore the effectiveness of employing the venerable VGG16 model, a stalwart in the realm of CNN architectures, for the noble cause of advanced glaucoma detection. Through the strategic application of transfer learning and fine-tuning methodologies, the endeavor seeks to imbue the VGG16 model with the prowess to discern varying degrees of glaucomatous damage within a diverse dataset of retinal images. The core of this project pulsates with the desire to scrutinize the model's accuracy, sensitivity, and specificity in unraveling the intricate tapestry of glaucoma progression. Through meticulous experimentation and rigorous evaluation, it endeavors to illuminate the path towards a future where even the faintest whispers of glaucoma do not escape the vigilant gaze of modern technology. The stakes are high, for the ramifications of success extend far beyond the confines of the laboratory, carrying the potential to revolutionize early diagnosis and intervention strategies, thus alleviating the burden of vision impairment inflicted by glaucoma. At the heart of this endeavor lies the exploration of Convolutional Neural Networks (CNNs) as the vanguard in the battle against glaucoma, with the esteemed VGG16 architecture leading the charge. Armed with a trove of retinal fundus images meticulously pre-processed for analysis, the project sets out to harness the formidable capabilities of the VGG16 model in the realm of glaucoma classification. The mission is clear: to empower the model with the ability to discern the telltale signs of glaucoma within the intricate patterns and structures adorning the retinal canvas. The potency of the VGG16 model lies in its innate ability to sift through the visual cacophony of retinal images, extracting salient features that serve as harbingers of glaucomatous affliction. Through the judicious exploitation of these features, the model holds the promise of distinguishing between the serenity of ocular health and the insidious encroachment of glaucoma. This project stands as a beacon illuminating the path towards the development of sophisticated diagnostic tools tailored specifically for glaucoma detection, with VGG16 serving as the cornerstone of this technological crusade. The realization of VGG16's potential in the realm of glaucoma classification heralds a new era in the battle against this silent scourge, where early diagnosis becomes not merely a possibility but
blindness worldwide, necessitates early detection for effective management and prevention of vision loss. In recent years, Convolutional Neural Networks (CNNs) have shown promise in automating glaucoma detection through analysis of retinal images. This project investigates the efficacy of employing the VGG16 model, a renowned CNN architecture, for advanced glaucoma detection. Leveraging transfer learning and fine-tuning techniques, the project aims to train the VGG16 model on a dataset comprising retinal images with varying degrees of glaucomatous damage. Through extensive experimentation and evaluation, the project endeavors to assess the model's accuracy, sensitivity, and specificity in identifying subtle signs of glaucoma progression. The outcomes of this project hold significant implications for enhancing early diagnosis and intervention strategies, potentially mitigating the burden of glaucoma-related vision impairment. This project explores the potential of Convolutional Neural Networks (CNNs), particularly the VGG16 architecture, for advanced glaucoma detection using retinal fundus images. The proposed method utilizes pre-processed fundus images to train the VGG16 model for accurate glaucoma classification. After training, the model can analyse new fundus images and predict the presence or absence of glaucoma. VGG16's strength lies in its ability to automatically extract relevant features from retinal images that might be indicative of glaucoma. By leveraging these features, the model can potentially differentiate healthy eyes from glaucomatous ones. This project aims to contribute to the development of efficient diagnostic tools for glaucoma detection using VGG16. The successful implementation of VGG16 for glaucoma classification has the potential to improve early diagnosis and patient outcomes. Key Words: Glaucoma, Deep Learning, Convolutional Neural Network (CNN), VGG16, Fundus Images, Image Classification
1.INTRODUCTION Glaucoma stands as a formidable adversary in the realm of eye diseases, reigning as a leading culprit behind irreversible blindness across the globe. The urgency for early detection becomes paramount in combating its progression and salvaging precious vision. In recent times, the emergence of Convolutional Neural Networks (CNNs)
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 393