Abstract: One of the most difficult challenges in medical image processing is detecting brain tumours. The challenge is challenging to complete since the photographs have a lot of variety, as brain tumours exist in a variety of shapes and textures. Brain tumours are made up of several types of cells, and the cells can reveal information about the tumor's nature, severity, and rarity. Tumors can appear in a variety of areas, and the location of a tumour can reveal information about the sort of cells that are creating it, which can help with further diagnosis. The process of detecting brain tumours can be made more difficult by issues that can be found in practically all digital images, such as lighting issues. The picture intensities of tumour and non-tumor images can overlap, making it challenging for any model to make accurate predictions from raw images. This research offers a novel method for detecting brain cancers from various brain pictures by using image preprocessing techniques such as histogram equalisation and opening, followed by a convolutional neural network. Apart from the picture preprocessing approaches that have been finalised for training, the study also explores the impact of alternative image preprocessing techniques on our dataset. The experiment was done out on a dataset that included tumours of various shapes, sizes, textures, and locations. For the classification challenge, a Convolutional Neural Network (CNN) was used. CNN achieved a recall of 98.55 percent on the training set and 99.73 percent on the validation set in our research, which is quite impressive.
Keywords: Brain Tumor Detection, Convolutional Neural Networks, Deep Learning, Image Processing, Computer Vision, Computer-aided Diagnosis, Transfer Learning.
|
DOI:
10.17148/IJARCCE.2022.115203
