Abstract When it comes to the field of medical image processing, the classification of brain tumours is one of the most significant and difficult problems to solve. As a result of the fact that manual classification with the assistance of humans might result in incorrect diagnoses and forecasts. In addition to this, whenever there is a substantial amount of information that must be processed manually, the process develops into a lengthy activity that is difficult to complete. As a result of the fact that brain tumours can take on a wide variety of forms, as well as the fact that there is a certain degree of similarity among normal and tumor tissues, it can be challenging to distinguish sections of a patient's brain that contain tumours from scans of that brain. As a result, a model is constructed to detect brain tumours from 2D magnetic resonance images of the brain by utilising a hybrid deep learning technique. This methodology is then accompanied with both traditional classification techniques and deep learning approaches. The application of the concept in clinical settings is the ultimate goal. The research was carried out using a Kaggle and BRaTS MICCAI dataset that had a wide range of tumours, each of which had its own size, location, and form, in addition to differing levels of image intensity. A total of 6 various classification methods namely Support Vector Machine (SVM), K-Nearest Neighbor (KNN), Multi-layer Perceptron (MLP), Logistic Regression (LR), and Naive Bayes (NB) were used when doing the conventional phase of categorization. When compared to these conventional classifications models, the SVM produced the most accurate results. After that, a Convolutional Neural Network (CNN) is used, which, when compared to the traditional classifiers, shows a significant enhancement in overall performance. Various Layers of CNN using different split ratio of dataset was evaluated. It is observed from the experimental findings that 5 layered CNN can obtain the highest performance accuracy of 97.86% using 80:20 split ratio.
Alan : Mühendislik
Dergi Türü : Uluslararası
Benzer Makaleler | Yazar | # |
---|
Makale | Yazar | # |
---|