An improved extreme learning machine for pathological brain detection

In the past decade, pathological brain detection has made remarkable progress, as a result many successful pathological brain detection systems (PBDSs) have been developed. However, the accuracy of these systems still needs significant improvement in order to meet the necessity of real world diagnostic situations. In this paper, an efficient PBDS based on MR images is proposed that markedly improves the recent results. The proposed system makes use of contrast limited adaptive histogram equalization (CLAHE) to enhance the quality of the input MR images. Thereafter, fast discrete curvelet transform via wrapping (FDCT-WR) strategy is harnessed to extract curve like features from the preprocessed images. Subsequently, the dimensionality of the features is reduced using PCA+LDA approach. Finally, a novel improved learning algorithm called IHPSO-ELM is proposed that combines self-organizing hierarchical PSO (HPSO) and extreme learning machine (ELM) for segregation of MR images as pathological or healthy. The HPSO is utilized to optimize the input weights and hidden biases of single-hidden-layer feed-forward neural networks (SLFN), whereas one analytical method is used for determining the output weights. The proposed algorithm performs optimization according to both the root mean squared error (RMSE) and norm of the output weights of SLFNs. Extensive experiments are carried out using three benchmark datasets and the results are compared against other competent schemes. The experimental results indicate that the proposed scheme offers better performance compared to its counterparts. Further, it has been noticed that the proposed IHPSO-ELM obtains higher accuracy and compact network architecture than conventional learning methods.