Brain Tumour Detection Using MRI Images and CNN Architecture

• Published in: 2024 International Conference on Communication, Computer Sciences and Engineering (IC3SE) pp. 540 - 548
• Main Authors: Kushwaha, Pradeep Kumar, Rana, Ajay, Lohani, Bhanu Prakash, Gupta, Amardeep, Laishram, Clinton, Ayyasamy, Kishor
• Format: Conference Proceeding
• Language: English
• Published: IEEE 09.05.2024
• Subjects: Accuracy; Brain tumour; Confusion Matrix; Convolutional neural network (CNN); Convolutional neural networks; Dataset; Instruments; Loss function; Machine Learning; Machine learning algorithms; Magnetic resonance imaging; Medical services; MRI images; Prognostics and health management; Sensitivity; Training; Tumors
• DOI: 10.1109/IC3SE62002.2024.10593571

The brain tumour is a phenomenon involving the creation of cells which multiply within the brain and they are different from the normal cells. Among all platforms, magnetic resonance imaging (MRI) has the highest level of accuracy in finding these brain cancer cells. Magnetic resonance imaging or MRI has made it possible to visualise tissues closely and then the diagnosis can be made. The result of an MRI scan will help in detecting the presence of a brain tumour or determining that there isn't a tumour. In the past few year's algorithms like deep learning and machine learning driven by AI have improved computer-aided image analysis tools which can now achieve the sensitivity of top radiologists. By modernising the diagnostic systems, the speed of tumour detection and error rate can be boosted which both play an essential role in a successful cancer treatment. This, in turn, lowers the possibility of healthcare providers misdiagnosing the patient and supports them in the right purposeful treatment. Here we have a research application where a convolutional neural network (CNN) discriminates brain tumour images compared to others and its implementation is carefully presented. The primary objective of this research is to employ Convolutional Neural Networks (CNN) as a machine learning technique to facilitate the detection and classification of brain tumours. However, the performance of the untrained and pre- trained CNN manifests through a precision of 95% and classification accuracy rate for training and testing respectively. Hence, such data provide the strongest evidence that in the situation of brain tumour prognosis, CNN is the most important instrument.