High Impact Factor : 4.396 icon | Submit Manuscript Online icon |

Performance Comparison of FMRI Image Analysis Techniques


Manu Chauhan , PEC university of technology; Bipan Kaushal, PEC university of technology


FMRI, CNN, SVM, Neuroimaging, Feature extraction, Accuracy


Functional magnetic resonance imaging (FMRI) is a neuroimaging process that utilizes MRI technique for measuring brain activity by the detection of associated blow flow change. Numerous methods for FMRI image analysis has been continuously evolving day by day. The techniques such as state vector machines (SVM) and linear neural networks are very common for the analysis of FMRI images. However, continuous efforts are being placed to improve the accuracy levels. So in this paper a new method for the classification of MRI images is proposed to meet the requirement. Convolution neural network (CNN) is a technique which is currently being used in the field of image processing. In this thesis work CNN along with k-means clustering, feature extraction and nuclei segmentation are implemented for the efficient detection of activated image features from a given set of MRI database. A set of 40 MRI images is collected and segmented by the technique known as nuclei segmentation. The noise is removed by cropping the non-brain portion from the images. Further, Feature extraction and k-mean clustering algorithm is implemented in order to extract features and partition them into the clusters. Finally, the classification is done by the methods including SVM, linear neural network and convolution neural network. The results are obtained in terms of accuracies and are compared with each other.

Other Details

Paper ID: IJSRDV4I50045
Published in: Volume : 4, Issue : 5
Publication Date: 01/08/2016
Page(s): 20-22

Article Preview

Download Article