Novel BDefRCNLSTM: an efficient ensemble deep learning approaches for enhanced brain tumor detection and...

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Computational Biology

Background:

• Manual brain tumour identification from MRI is time-consuming and error-prone.
• Automated solutions are needed to improve accuracy and efficiency in brain tumour diagnosis.
• Complex tumour characteristics require advanced analytical methods.

Purpose of the Study:

• To introduce a novel ensemble deep learning model, BDefRCNLSTM, for brain tumour classification and segmentation.
• To enhance feature extraction and selection for improved diagnostic accuracy.
• To provide a clinically viable automated solution for brain tumour detection.

Main Methods:

• Developed a boosted deformable and residual convolutional network with bi-directional convolutional long short-term memory (BDefRCNLSTM).
• Integrated entropy-based local binary pattern (ELBP) for spatial semantic feature extraction.
• Employed enhanced sooty tern optimisation (ESTO) for optimal feature selection and an improved X-Net for segmentation.

Main Results:

• The BDefRCNLSTM model achieved over 99% accuracy in both brain tumour classification and segmentation.
• The proposed model outperformed existing state-of-the-art approaches on multiple datasets.
• Demonstrated the effectiveness of integrated feature selection and segmentation techniques.

Conclusions:

• The BDefRCNLSTM model represents a clinically viable solution for automated brain tumour diagnosis.
• The approach can assist radiologists in making faster and more reliable decisions.
• Optimised feature selection and advanced segmentation significantly improve diagnostic accuracy.