Computational Intelligence and Metaheuristic Techniques for Brain Tumor Detection through IoMT-Enabled MRI Devices

Table 1

The related work of the machine learning approaches for brain tumor.

Authors	Dataset	Access	Technique	Limitations
Manogaran, G. et al. [40]	MRI dataset	Open	Thresholding, gamma distribution	The proposed work needed to accelerate real-time medical applications and computation time.
Patil, D et al. [36]	BRATS 2015	Open	Local binary pattern, empirical wavelet transform, dynamic fuzzy histogram equalization	The work lacked the accuracy of interpretation.
Arun, N et al. [34]	MRI dataset	Open	Artificial neural network, machine learning	The technique affected the feature extraction because of difficulty in segmentation.
Nazir, M et al. [35]	Harvard dataset	Open	K mean clustering, discrete cosine transform	The proposed study is incapable of correctly classifying malignant brain tissues.
Garg, G et al. [46]	MRI dataset	Open	Principal component analysis, gray level cooccurrence matrix, stationary wavelet transform, Otsu’s threshold	The technique needed large dataset for training, had high time complexity, did not work for small dataset and required expensive GPUs which ultimately increased cost to the users.
Kumar, M et al. [21]	MRI dataset	Open	Balance contrast enhancement technique, canny operator, fuzzy c-means	The technique needed to be performed on real medical images of patients in order to solve urgent diagnostic problems of patients.
Bahadure, N et al. [41]	MRI dataset	Open	Berkeley wavelet transformation, support vector machine	The methodology required the combination of more than one classifier and feature selection techniques to improve the accuracy.