Neuronal feature extraction through autonomous segmentation using density conscious artificial immune algorithm

Detection of geometric features in images of neuronal structures is important step in devising computational tools for diagnosis and cure of neuro-psychotic diseases. Examples of such structures include those in the hippocampal regions of the brain, and detection of nerve-fibre streamline bundles in Diffusion-Weighted Magnetic Resonance Imaging (DW-MRT) of brain. Autonomous detection of the neuronal structures therefore facilitates exploration of the correspondence, which can further be used to diagnose, predict neuro-psychotic disorders and also to prescribe timely cures for such diseases. Intricate structural relations are often missed out during visual analysis, and effective computational algorithms employing statistical and geometric models are required. The current paper investigates two density conscious models based on artificial immune system, namely AIDEN, and DCAIGMM for autonomous detection of the arc-features. The effectiveness of the model in presence of noisy brain images has been investigated.