Deep Learning Models for Brain Cancer Segmentation and Prognostic Analysis / Hiba Djari

Public ISBD Titre : Deep Learning Models for Brain Cancer Segmentation and Prognostic Analysis Type de document : document électronique Auteurs : Hiba Djari ; Malak Lamara, Auteur ; Hamdi ,Skander, Directeur de thèse Editeur : Setif:UFA Année de publication : 2025 Importance : 1 vol (64 f .) Format : 29 cm Langues : Anglais (eng) Catégories : Thèses & Mémoires:Informatique Mots-clés : Informatique Index. décimale : 004 Informatique Résumé : Deep Learning (DL) has introduced transformative potential in medical diagnostics, particularly in oncology, where complex tumor structures present significant challenges for treatment planning. The accurate segmentation of brain tumors from Magnetic Resonance Imaging (MRI) is a critical first step, and recent advancements in DL have greatly enhanced the efficiency and reliability of this process. By integrating these precise segmentation outputs with clinical and genetic data, prognostic analysis can be substantially improved, paving the way for more accurate tumor prediction and robust clinical decision support. This thesis aims to develop and evaluate a comprehensive pipeline for brain tumor analysis, from segmentation to survival prediction. It focuses on comparing advanced DL architectures for segmenting brain tumors and explores the adaptation of large-scale, general-purpose vision models for this specialized medical task. The ultimate goal is to fuse imaging features with clinical and genetic data to accurately estimate patient survival probability and stratify risk levels, thereby enhancing personalized treatment strategies. This work offers a thorough exploration of both established and cutting-edge DL models for brain tumor segmentation and prognosis. The first contribution is a comparative analysis of segmentation models, including a standard U-Net, a DeepResUNet, and a VGG19-based U-Net, alongside foundational models like a fine-tuned Segment Anything Model (SAM) and the specialized Medical Segment Anything Model (Med- SAM). The results underscored the effectiveness of MedSAM, which demonstrated superior performance . The second contribution introduces a prognostic analysis pipeline that leverages features extracted from the segmentation masks. This pipeline proved highly effective, with the Support Vector Machine (SVM) model showing notable success in accurately identifying high-risk patients, while the XGBoost model demonstrated strong predictive power for classifying low-risk patients. Overall, these contributions validate a powerful, integrated approach for brain cancer analysis. The findings confirm that leveraging both CNN-based architectures and fine-tuned foundational models can yield highly accurate tumor delineations. Furthermore, the successful integration of these segmentation results into a prognostic pipeline demonstrates the critical role of automated image analysis in fostering personalized and more effective clinical management of brain tumors. Note de contenu : Sommaire 1 Introduction 14 1.1 Research Context and Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.2 Research Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.3 Structure of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2 Background 17 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2 Deep Learning Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.1 CNN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.2 U-Net . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.2.3 Deep ResUNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.2.4 VGG19-UNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.2.5 SAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.2.6 MedSAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.3 Standard Classifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.3.1 SVM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.3.2 Logistic Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.4 Ensemble Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.4.1 Random Forest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.4.2 XGBoost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.4.3 CatBoost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.4.4 Majority voting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3 Literature review 26 3.1 Overview on Brain Cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 4 Methodology 32 4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.2 Dataset Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.3 Models Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.3.1 Data Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.3.2 U-Net . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.3.3 Deep ResUNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.3.4 VGG19-UNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.3.5 SAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.3.6 MedSAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.4 Prognostic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.4.1 Model and Dataset Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.4.2 Tumor Segmentation and Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . 47 4.4.3 Feature Vector Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.4.4 Prognostic Prediction and Thresholding . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4.5 Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4.6 Feature Extraction from Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4.7 Classifier Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4.8 Confusion matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 5 General Conclusion Côte titre : MAI/0964 Deep Learning Models for Brain Cancer Segmentation and Prognostic Analysis [document électronique] / Hiba Djari ; Malak Lamara, Auteur ; Hamdi ,Skander, Directeur de thèse . - [S.l.] : Setif:UFA, 2025 . - 1 vol (64 f .) ; 29 cm. Langues : Anglais (eng) Catégories : Thèses & Mémoires:Informatique Mots-clés : Informatique Index. décimale : 004 Informatique Résumé : Deep Learning (DL) has introduced transformative potential in medical diagnostics, particularly in oncology, where complex tumor structures present significant challenges for treatment planning. The accurate segmentation of brain tumors from Magnetic Resonance Imaging (MRI) is a critical first step, and recent advancements in DL have greatly enhanced the efficiency and reliability of this process. By integrating these precise segmentation outputs with clinical and genetic data, prognostic analysis can be substantially improved, paving the way for more accurate tumor prediction and robust clinical decision support. This thesis aims to develop and evaluate a comprehensive pipeline for brain tumor analysis, from segmentation to survival prediction. It focuses on comparing advanced DL architectures for segmenting brain tumors and explores the adaptation of large-scale, general-purpose vision models for this specialized medical task. The ultimate goal is to fuse imaging features with clinical and genetic data to accurately estimate patient survival probability and stratify risk levels, thereby enhancing personalized treatment strategies. This work offers a thorough exploration of both established and cutting-edge DL models for brain tumor segmentation and prognosis. The first contribution is a comparative analysis of segmentation models, including a standard U-Net, a DeepResUNet, and a VGG19-based U-Net, alongside foundational models like a fine-tuned Segment Anything Model (SAM) and the specialized Medical Segment Anything Model (Med- SAM). The results underscored the effectiveness of MedSAM, which demonstrated superior performance . The second contribution introduces a prognostic analysis pipeline that leverages features extracted from the segmentation masks. This pipeline proved highly effective, with the Support Vector Machine (SVM) model showing notable success in accurately identifying high-risk patients, while the XGBoost model demonstrated strong predictive power for classifying low-risk patients. Overall, these contributions validate a powerful, integrated approach for brain cancer analysis. The findings confirm that leveraging both CNN-based architectures and fine-tuned foundational models can yield highly accurate tumor delineations. Furthermore, the successful integration of these segmentation results into a prognostic pipeline demonstrates the critical role of automated image analysis in fostering personalized and more effective clinical management of brain tumors. Note de contenu : Sommaire 1 Introduction 14 1.1 Research Context and Motivation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.2 Research Contributions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 1.3 Structure of thesis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 2 Background 17 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2 Deep Learning Approaches . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.1 CNN . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 2.2.2 U-Net . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18 2.2.3 Deep ResUNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19 2.2.4 VGG19-UNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.2.5 SAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 2.2.6 MedSAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.3 Standard Classifiers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.3.1 SVM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 2.3.2 Logistic Regression . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.4 Ensemble Learning . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.4.1 Random Forest . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23 2.4.2 XGBoost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.4.3 CatBoost . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.4.4 Majority voting . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 24 2.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 25 3 Literature review 26 3.1 Overview on Brain Cancer . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 26 3.2 Related Works . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 3.3 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 4 Methodology 32 4.1 Overview . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.2 Dataset Description . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32 4.3 Models Implementation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.3.1 Data Preparation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.3.2 U-Net . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 33 4.3.3 Deep ResUNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35 4.3.4 VGG19-UNet . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37 4.3.5 SAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39 4.3.6 MedSAM . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 44 4.4 Prognostic Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46 4.4.1 Model and Dataset Initialization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.4.2 Tumor Segmentation and Feature Extraction . . . . . . . . . . . . . . . . . . . . . . . 47 4.4.3 Feature Vector Construction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47 4.4.4 Prognostic Prediction and Thresholding . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4.5 Visualization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4.6 Feature Extraction from Segmentation . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4.7 Classifier Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 48 4.4.8 Confusion matrices . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55 4.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 5 General Conclusion Côte titre : MAI/0964

Exemplaires (1)

Code-barres	Cote	Support	Localisation	Section	Disponibilité
MAI/0964	MAI/0964	Mémoire	Bibliothèque des sciences	Anglais	Disponible Disponible

---

1 (1 - 1 / 1)