• Version
• Download 15
• File Size 527.07 KB
• File Count 1
• Create Date 4 July 2023
• Last Updated 4 July 2023

A Framework for Classifying Brain Tumours using Convolution Neural Networks and Open-Source Computer Vision

Amrut Ranjan Jena1, Madhusmita Mishra2, Ritam Ghosh3, Labanya Nayak4, Arkodeep Dey5

1Department of CSE & GNIT, Kolkata

2Department of CSE & DSCSITSC, Kolkata

3,4,5Department of CSE & GNIT, Kolkata

---------------------------------------------------------------------***---------------------------------------------------------------------

Abstract -A doctor diagnoses a patient correctly using the clinical records before prescribing the right medication. The diagnosis of the magnetic resonance imaging report to identify the brain tumour can occasionally take a long time for a medical professional. It has been noted that the tumour tissues resemble normal tissues in appearance. Marking the tumour area in the brain is a difficult task because of this. Cancer is caused by tumours. In order to avoid cancer, tumours must be detected at an early stage and the patient must receive the appropriate treatment. The creation of an automated method to identify and categorise the different types of tumours found in the human brain would simplify this difficult task. In order to detect and categorise tumour and no tumour in the human brain, a model is proposed in this paper by fusing open-source computer vision with three different types of convolution neural networks architectures. The residual neural network design, out of the three alternative hybridised CNN architectures, provides the highest accuracy, with 98.52% and 99.73% over testing and training data, respectively.

Key Words: Brain tumour, CNN, OpenCV, MRI, Cancer, Hybridization