brain tumors classification based on segmentation techniques and wavelet transform

This paper aims to provide better approaches for segmenting and classifying brain tumours using magnetic resonance images (mri). in this paper, the wavelet features are formed by the transformation of probability density function (pdf) to spectrogram images using continuous wavelet transform (2d-cwt), which is a simple method for extracting features, whereas the feature extraction methods (pdf and 2d-cwt) are improving the performance. in addition, a morphological operation for segmenting images and a convolutional neural network (cnn) are utilized as a classifier in order to increase the segmentation performance. on the brats2019 dataset, the method’s performance is assessed in terms of f1-score and tumour region segmentation accuracy. this achieved the greatest results, with accuracy and f1-score of 97.37 % and 97.43 %, respectively.