Deep Ensemble Model for Brain Age Prediction in MRI with Hybrid Optimal Feature Selection

Deep learning (DL) has a tremendous deal of potential for accurate brain age prediction using neuroimaging data, but its performance is frequently limited by the size of the training dataset and the computer’s memory needs. Under this circumstance, a unique brain age prediction model with the following five phases of development. The input image is first pre-processed via median filtering, which keeps the edges of the raw image while removing the noise. The pre-processed images are segmented using the improved Balanced Iterative Reducing and Clustering employing Hierarchies (BIRCH) algorithm. Then, features including statistics (Mean, Median, Standard Deviation), Improved Median Robust Extended LBP (MRELBP), and sharpness score are extracted from these segmented images. The suggested coot interfaced Archimedes with Gaussian Map Estimation (CIAGME) optimization approach, which combines the Archimedes and coot optimization techniques, will select the optimal features from those extracted features. The deep ensemble approach, which includes deep classifiers like CNN, Bi-LSTM, and Deep maxout, will provide a prediction based on the optimal features that have been chosen. Lastly, the recommended CIAGME-based predictions model’s outperformance is evaluated, and the result is effectively evaluated.