Optimal Trained Ensemble of Classifiers for Brain Age Prediction Using MRI

• Published in: 2024 IEEE International Conference on Information Technology, Electronics and Intelligent Communication Systems (ICITEICS) pp. 1 - 9
• Main Authors: G.S, Vishnupriya, Rajakumari, Brintha S.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 28.06.2024
• Subjects: brain age prediction; Brain modeling; DBSCAN; Deepmaxout; Feature extraction; Filtering; GRU; Image segmentation; Magnetic resonance imaging; MRI; Noise; Predictive models; RNN; Self Improved Bald Eagle Search Optimization
• DOI: 10.1109/ICITEICS61368.2024.10625565

Prediction of brain age is necessary for a better understanding of the development of the human brain and ageing. Recently MRI data have been used for the estimation of brain age. However, variance in the development of brain age between the different as well as similar subjects gives variations in the developments at different ages. Therefore, advanced models are needed to predict brain age more accurately. Hence, this paper intends to propose an optimally trained ensemble classifier for brain age prediction by using MRI data. Initially, the MRI image is preprocessed by using the Bilateral Filtering process to eliminate the unwanted noise. Subsequently, the preprocessed images are segmented by the DBSCAN model and followed by feature extraction. Features such as LGBPHS, statistical features and pixel features are extracted from the segmented image. Finally, the prediction process is done via an ensemble classification model, which is a combination of RNN, Deepmaxout and GRU. Moreover, the weights of all the classifiers are optimally tuned by a proposed Self Improved-Bald Eagle Search Optimization (SI-BESO) to enhance its performance. The performance of the suggested ensemble model is evaluated over the conventional algorithms as well as classifiers in terms of performance measures like MAPE, MSE, MSLE, and MAE.