International Research Journal of Engineering and Technology (IRJET)
e-ISSN: 2395-0056
Volume: 11 Issue: 04 | Apr 2024
p-ISSN: 2395-0072
www.irjet.net
Brain Tumor Grade Classification in MR images using Deep Learning 1Shruti Kolte, 2Satyajeet Pathare, 3Pranay Sune, 4Sudhanshu Gomase, 5Harshal Gothe 1Professor, 2,3,4,5Students
Department of Computer Science and Engineering, Priyadarshini JL College of Engineering, Nagpur, India --------------------------------------------------------------------------***--------------------------------------------------------------------------ABSTRACT Brain tumour and hemorrhages are critical brain diseases that require accurate and early diagnosis for effective treatment and management. This work proposes a hybrid deep learning and traditional machine learning approach for automated detection and diagnosis of brain tumours and hemorrhages using magnetic resonance imaging (MRI) scans. The developed system employs a combination of deep convolutional neural networks and engineered feature-based classifiers to leverage the representation learning capabilities of deep learning and the interpretability of traditional models. Expert subject knowledge is provided by hand-crafted features, and the deep learning models directly acquire hierarchical feature representations from the MRI images.. A fusion of predictions from both models is used to improve diagnostic accuracy. The system was trained and evaluated on a dataset of 3000 MRI scans categorized by tumour type and hemorrhage presence. Results demonstrate that the hybrid system outperforms either individual approach with 92% accuracy for tumour classification and 94% accuracy for hemorrhage detection. The integrated system provides accurate, automatic detection of critical brain disorders using MRI scans to assist healthcare professionals in early diagnosis and treatment planning. This work demonstrates the potential of hybrid AI systems for improving computer-aided diagnosis in healthcare. Keywords: Deep learning, convolutional neural networks, machine learning, radiology, brain tumor detection, hemorrhage detection, magnetic resonance imaging
I.
INTRODUCTION
Brain tumours and hemorrhages are critical medical conditions that can be life-threatening if not detected and diagnosed accurately in the early stages. However, the accurate and timely diagnosis of these brain disorders remains a key challenge in healthcare. Medical imaging techniques such as magnetic resonance imaging (MRI) and computed tomography (CT) scans are vital for non-invasive screening and detection of abnormalities in the brain. However, manually analyzing the large volume of scans to identify tumours, hemorrhages, and other neurological conditions can be error prone, time-consuming, and dependent on radiologist expertise. This highlights the need for automated computer-aided diagnosis (CAD) systems that can rapidly and reliably analyze medical images to detect brain disorders. Recent advances in deep learning, especially convolutional neural networks (CNNs), have shown immense potential for medical image analysis and precision diagnosis. CNNs are specialized deep neural networks which exploit the 2D structure of images through convolution operations and hierarchical feature learning. In contrast to traditional CAD systems relying on hand-crafted features, CNNs can automatically learn discriminative features directly from medical image data to detect abnormalities and classify pathologies. State-of the-art CNN architectures like Res Net and Dense Net have achieved high performance on tumour classification and brain disorder prediction using MRI and CT scans. However, most deep learning techniques act as black-box models, lacking interpretability and reliance on large labelled datasets which are often limited in healthcare. Brain cancer is a highly serious illness that kills a lot of people. To enable early diagnosis, a technique for detecting and classifying brain tumors is available. One of the most difficult challenges in clinical diagnostics is classifying cancer.
© 2024, IRJET
|
Impact Factor value: 8.226
|
ISO 9001:2008 Certified Journal
|
Page 856