DELINEATION AND CLASSIFICATION OF BRAIN TUMORUSING MRI SEQUENCES

Abstract

According to a global cancer statistical report 2020, cancer is the world’s second cause of death, with the prediction of a record of 19.3 × 106 newer cases of cancer along with 10.0 × 106 deaths in 2020. Among all type of cancers, central nervous system (CNS) and brain cancer are to be top twelve leading cause of death/ and reported nearly 308,102 (1.7%), new brain and CNS cancer cases and 251,329 (2.54%), brain and CNS death in 2020. Brain tumors are one of the critical malignant neurological cancers with the highest number of deaths and injuries worldwide. They are categorized into two major classes, high-grade glioma (HGG) and low-grade glioma (LGG), with HGG being more aggressive and malignant, whereas LGG tumors are less aggressive, but if left untreated, they get converted to HGG. Thus, the classification of brain tumors into the corresponding grade is a crucial task, especially for making decisions related to treatment. The work presented in the thesis is divided into eight chapters. Chapter 1 presents an introduction to medical imaging, while an in-depth literature survey is conducted in chapter 2, gives a detailed overview of the current state-of-the-art techniques for the classification/segmentation of brain tumors using publicly available datasets. Also, this chapter gives a detailed overview of datasets (i.e., BraTS, Figshare Brain Tumor MRI scan) and their Pre-processing used for experiment purposes. Chapter 3 presents a machine intelligence framework, namely GRGE, based on radiomics features extraction to discriminate between HGG and LGG using the BraTS i dataset. The BraTS dataset is divided using the five-fold cross-validation (CV) scheme in a stratified manner and utilized the training set of each fold to select the best and the most significant features for that fold using a binary genetic algorithm. The extracted and selected radiomics features are normalized using Z-score normalization. The normalized set of features then passed to the different machine learning classifiers: LR, SVM, kNN, and Extra Randomized Tree (ERT), where ERT achieved the best performance with an Acc of 95.44%, Sens of 98.57%, and Spec of 86.67%. Chapter 4 presents a computational decision support system, CGHF, for the grade of glioma classification more accurately and in a non-invasive manner using hybrid radiomics and stationary wavelet-based features. The training dataset of Brain Tumor Segmentation (BraTS) Challenge 2018 is used for performance evaluation, and calcu lation is done based on the radiomics features for three different regions of interest. The classifier, Random Forest, is trained on these features and predicted the grade of glioma. At last, validated the performance by using the five-fold CV scheme. The state-of-the-art performance is achieved considering metric ⟨Acc, Sens, Spec, Score, MCC, AUC⟩ ≡⟨97.54%, 97.62%, 97.33%, 98.3%, 94.12%, 97.48%⟩ with machine learning predictive model, Random Forest (RF), for brain tumor patients’ classification. Chapter 5 presents a novel framework, IBRDM, which uses discrete wavelet transform (DWT) based fusion of MRI sequences and Radiomics feature extraction for brain tumor classification. We utilized the BraTS 2018 challenge training dataset for the performance evaluation of our approach and extract features from three regions of interest derived using a combination of several tumor regions. We used wrapper method-based feature selection techniques for selecting a significant set of features and utilize various machine learning classifiers, RF, Decision Tree (DT), and ERT, for training the model. We achieved state-of-the-art performance considering several performance metrics ⟨Acc, Sens, Spec, F1-score, MCC, AUC⟩ ≡⟨98.60%, 99.05%, 97.33%, 99.05%, 96.42%, 98.19%⟩ using five-fold CV scheme. Chapter 6 presents an automated computer-aided diagnosis (CAD) system, DT CoMBTC,for effective brain tumor classification to classify into three different categories, such as glioma tumor (GT), meningioma tumor (MT), and pituitary tumor (PT). The proposed methodology, DT-CoMBTC, utilizes the dual-tree complex wavelet transform (DT-CWT) and center of mass (CoM) to pre-processing images and train the CNN ii predictive models using publicly available Figshare brain tumor dataset. To check the proposed system’s robustness, DT-CoMBTC is validated in a two-way: (i) hold-out and (ii) five-fold cross-validation scheme. Experimentation findings demonstrate that DT-CoMBTC has performed well over many recent baseline approaches in terms of Acc, Sens, and Spec. Chapter 7 focuses on brain tumor segmentation by using the different normaliza tion techniques for pre-processing brain tumor MRI datasets. In the proposed work, CBSN, we consider well-known normalization techniques such as Gaussian mixture models (GMM), Fuzzy C-means (FCM), and Z-score normalization for pre-processing the brain tumor BraTS 2018 dataset. We utilized three variants of U-Net architecture, convolutional block attention module (CBAM), squeeze and excitation module (SEM), and refinement module (RM), for the segmentation of the regions of interest (ROIs). Utilizing Z-score normalization performs better than other normalization techniques for tumor core (TC) and whole tumor (WT) segmentation. In contrast, FCM per forms superior to the other two normalization techniques on enhancement tumor (ET) segmentation. Finally, chapter 8 concludes the work done in all the above chapters by highlighting the advantages, limitations, and suggestions for future scope research work.