University Sétif 1 FERHAT ABBAS Faculty of Sciences
Détail de l'auteur
Auteur Tarek Laouamri |
Documents disponibles écrits par cet auteur
Ajouter le résultat dans votre panier Affiner la recherche
Hybridization of Deep and Generative Models for Accurate Generation and Classification Chest X-ray Abnormalities / Mohamed Chakib Zaghlaoui
Titre : Hybridization of Deep and Generative Models for Accurate Generation and Classification Chest X-ray Abnormalities Type de document : texte imprimé Auteurs : Mohamed Chakib Zaghlaoui, Auteur ; Tarek Laouamri ; Abdelouahab Moussaoui, Directeur de thèse Editeur : Sétif:UFS Année de publication : 2023 Importance : 1 vol (66 f.) Format : 29 cm Langues : Anglais (eng) Catégories : Thèses & Mémoires:Informatique Mots-clés : Deep learning, generative models
Chest X-ray abnormalities
Classification
Augmentation
Convolutional neural networks
Generative
Adversarial networksIndex. décimale : 004 - Informatique Résumé : Medical imaging plays a crucial role in the diagnosis and treatment of various medical conditions. In recent years, deep learning
models have shown remarkable success in automating the analysis
of medical images, particularly in the field of chest X-ray examination. However, the accurate generation and classification of chest
X-ray abnormalities still pose significant challenges.
This master thesis aims to address these challenges by proposing
a novel hybrid approach that combines deep learning and generative
models. The primary objective is to leverage the strengths of both
model types to achieve more accurate and reliable results in the
generation and classification of chest X-ray abnormalities.
The proposed hybrid model architecture consists of a deep convolutional neural network (CNN) as the primary classifier, and a
generative adversarial network (GAN) for generating realistic and
diverse chest X-ray abnormality samples. By training the hybrid
model on a large dataset of labeled chest X-ray images, it learns to
generate synthetic abnormality samples that closely resemble real
abnormalities.
The generated abnormality samples are then used to augment the
original dataset, creating an augmented dataset with increased diversity and representation of abnormalities. This augmented dataset
is then utilized to retrain and fine-tune the deep CNN classifier, enhancing its ability to accurately classify chest X-ray abnormalities.
Experimental evaluations conducted on a comprehensive dataset
of chest X-ray images demonstrate the effectiveness of the proposed
hybrid model. The results indicate significant improvements in both
the generation of realistic abnormality samples and the accuracy
iv
of abnormality classification compared to traditional deep learning
approaches.
In conclusion, the hybridization of deep and generative models
presented in this thesis provides a promising avenue for enhancing the generation and classification of chest X-ray abnormalities.
The ability to generate diverse and realistic abnormality samples,
coupled with improved classification accuracy, can greatly benefit medical professionals in diagnosing and treating patients with
chest-related medical conditions.
Côte titre : MAI/0820
En ligne : https://drive.google.com/file/d/1aSvvc1K6fCXHxQLUAEct48J0zfamK7xO/view?usp=drive [...] Format de la ressource électronique : Hybridization of Deep and Generative Models for Accurate Generation and Classification Chest X-ray Abnormalities [texte imprimé] / Mohamed Chakib Zaghlaoui, Auteur ; Tarek Laouamri ; Abdelouahab Moussaoui, Directeur de thèse . - [S.l.] : Sétif:UFS, 2023 . - 1 vol (66 f.) ; 29 cm.
Langues : Anglais (eng)
Catégories : Thèses & Mémoires:Informatique Mots-clés : Deep learning, generative models
Chest X-ray abnormalities
Classification
Augmentation
Convolutional neural networks
Generative
Adversarial networksIndex. décimale : 004 - Informatique Résumé : Medical imaging plays a crucial role in the diagnosis and treatment of various medical conditions. In recent years, deep learning
models have shown remarkable success in automating the analysis
of medical images, particularly in the field of chest X-ray examination. However, the accurate generation and classification of chest
X-ray abnormalities still pose significant challenges.
This master thesis aims to address these challenges by proposing
a novel hybrid approach that combines deep learning and generative
models. The primary objective is to leverage the strengths of both
model types to achieve more accurate and reliable results in the
generation and classification of chest X-ray abnormalities.
The proposed hybrid model architecture consists of a deep convolutional neural network (CNN) as the primary classifier, and a
generative adversarial network (GAN) for generating realistic and
diverse chest X-ray abnormality samples. By training the hybrid
model on a large dataset of labeled chest X-ray images, it learns to
generate synthetic abnormality samples that closely resemble real
abnormalities.
The generated abnormality samples are then used to augment the
original dataset, creating an augmented dataset with increased diversity and representation of abnormalities. This augmented dataset
is then utilized to retrain and fine-tune the deep CNN classifier, enhancing its ability to accurately classify chest X-ray abnormalities.
Experimental evaluations conducted on a comprehensive dataset
of chest X-ray images demonstrate the effectiveness of the proposed
hybrid model. The results indicate significant improvements in both
the generation of realistic abnormality samples and the accuracy
iv
of abnormality classification compared to traditional deep learning
approaches.
In conclusion, the hybridization of deep and generative models
presented in this thesis provides a promising avenue for enhancing the generation and classification of chest X-ray abnormalities.
The ability to generate diverse and realistic abnormality samples,
coupled with improved classification accuracy, can greatly benefit medical professionals in diagnosing and treating patients with
chest-related medical conditions.
Côte titre : MAI/0820
En ligne : https://drive.google.com/file/d/1aSvvc1K6fCXHxQLUAEct48J0zfamK7xO/view?usp=drive [...] Format de la ressource électronique : Exemplaires (1)
Code-barres Cote Support Localisation Section Disponibilité MAI/0820 MAI/0820 Mémoire Bibliothéque des sciences Anglais Disponible
Disponible