| Titre : | Deep learning approach for early diabetic retinopathy diagnosis |
| Auteurs : | Kaouthar Manar Fellah, Auteur ; Samir Tigane, Auteur |
| Type de document : | Monographie imprimée |
| Editeur : | Biskra [Algérie] : Faculté des Sciences Exactes et des Sciences de la Nature et de la Vie, Université Mohamed Khider, 2022 |
| Format : | 1 vol. (56 p.) / couv. ill. en coul / 30 cm |
| Langues: | Anglais |
| Mots-clés: | Healthcare, Diabetic Retinopathy, CNN, Artificial Intelligence, Image Processing, Convolutional Neural Network |
| Résumé : |
Diabetes Mellitus (DM) is a metabolic illness that occurs when the body’s blood sugar levels become excessively high. It is a serious public health problem, affecting 463 million people worldwide and this number is projected to rise to 700 million by 2045. Diabetic Retinopathy (DR) is the most common specific complication of DM. DR is a leading cause of blindness among working-age adults. Early identification and treatment of DR can lower the risk of vision loss greatly. Since a manual diagnosis is prone to misdiagnosis and requires more effort, the automated methods for DR detection are cost and time effective. Deep learning has recently been one of the most popular strategies for improving experience in a range of fields, particularly medical image analysis and classifications. In this research, we demonstrate the use of convolutional neural networks (CNNs) on color fundus images. These images are preprocessed with various filters before being fed into the training model. Finally, experimental results show that the proposed approach outperforms similar works in the literature. |
| Sommaire : |
General introduction 1 1 State of the art 3 1.1 Introduction . 3 1.2 Artificial intelligence in healthcare 3 1.2.1 Clinical applications 4 1.3 Machine Learning 6 1.3.1 Supervised learning . 6 1.3.2 Unsupervised learning 6 1.3.3 Reinforcement learning 7 1.3.4 Semi-supervised learning 7 1.4 Deep Learning 8 1.4.1 Definition .. 8 1.4.2 Types of Deep Learning architectures 8 1.5 Artificial neural networks 9 1.5.1 Artificial neuron 9 1.5.2 Artificial neural network 9 1.6 Convolutional neural networks 10 1.6.1 Definition 10 1.6.2 Main operations of convolution 10 1.6.3 Types of layers in a CNN 12 1.6.4 Popular CNN architectures 15 1.7 Transfer learning 17 1.8 Diabetic retinopathy 17 1.8.1 The retina of the eye 17 1.8.2 Diabetes 18 1.8.3 Diabetic retinopathy 18 1.8.4 Fundus photography 22 1.8.5 Related works 22 1.9 Conclusion 24 I2 System design and implementation 25 2.1 Introduction 25 2.2 Global architecture of the system 25 2.2.1 Dataset 26 2.2.2 Preprocessing 27 2.2.3 Splitting dataset 29 2.2.4 CNN Learning 29 2.2.5 Prediction 30 2.2.6 Evaluation of a CNN model 31 2.3 Implementation of a Deep Learning architecture31 2.3.1 Frameworks , tools and libraries 31 2.3.2 Dataset preparation and preprocessing32 2.3.3 Building the CNN Model 36 2.4 Testing the CNN Model41 2.5 Conclusion 43 3 Experimentation and Results 44 3.1 Introduction 44 3.2 First proposed structure (Binary Classification) 44 3.2.1 The used Dataset 44 3.2.2 Specification of the used parameters 45 3.2.3 Results and discussion 45 3.3 Second Proposed Structure (Multiclass Classification) 48 3.3.1 The used Dataset 48 3.3.2 Specification of the used parameters. 49 3.3.3 Results and discussion 49 3.4 Comparison with other related works 52 3.5 Conclusion 54 Conclusion and Perspectives 55 Bibliography 5 |
Disponibilité (1)
| Cote | Support | Localisation | Statut |
|---|---|---|---|
| MINF/720 | Mémoire master | bibliothèque sciences exactes | Consultable |
