Mri Brain Deviations

MRI BRAIN IMAGES-A COMPARITIVE STUDY BETWEEN HGNN AND IPSONN

Dhanya G S#1

Dr. R. Joshua Samuel Raj#2, Sam Silva.A#3

1PG Scholar,

2Professor,

3Assistant Professor

#Department of Computer Science and Engineering, Anna University-Chennai,

Rajas Engineering College,

Tirunelveli, Tamilnadu, India [e-mail] Abstract— Image processing involves the management of images to remove information to highlight or suppress certain phases of the information, contained in the image or perform image analysis to extract hidden information. The recent imaging modalities in medicine, such as Magnetic Resonance Imaging (MRI) generate images directly in digital form. Estimation of the size of the whole organ, portions of the organ and/or objects surrounded

(ii) Parameters-the position of particle, its velocity, the learning parameters, own inertia, weight, and utmost amount of iterations are defined. (iii) Fitness (iv) Updating the velocity and position. (v) Stopping criteria, where the final optimal feature from the IPSO is exploited.

2.2.4 Tissue Classification by FFBNN: The feature set is given to the FFBNN classifier for training process and this classifier is represented as c-FFBNN.

Step1: Assign input weights to neurons

Step2: Calculate the learning error for the neural network [12].

.

3. Comparative Analysis

The classification performance of IPSONN and HGNN are analysed. HGNN utilised hybrid genetic approach and IPSONN utilized swarm optimization technique. This classification method result of IPSONN is shown in Table 1 and that of HGNN is shown in Table2.The graphical representation of the average performance of IPSONN and HGNN are shown in figures 2 a, b, c.

Table1: Performance of IPSONN method in classifying WM, GM, CSF, edema, tumor.

IPSONN WM GM CSF edema tumor

TP 1 1 1 1 0

FP 0 1 0 0 1

TN 4 3 4 4 4

FN 0 0 0 0 0

Sensitivity 100 100 100 100 0

FPR 0.0 25.0 0.0 0.0 20.0

ACC 100 80 100 100 80

Specificity 100 75 100 100

MRI brain images were utilized to analyse the results of the HGNN and IPSONN classification method. The performance analysis proved that the IPSONN method offers an average of 95%, 87%, 94% for accuracy, sensitivity and specificity measures, respectively [12]. Thus, the results show that the IPSONN achieved more classification performance than the HGNN method.

References

[1] J. N. Giedd, J. C. Rajapakse and J. L. Rapoport, “Statistical approach to segmentation of single-channel cerebral MR images,” IEEE Transactions on Medical Imaging, vol. 16, no. 2, pp. 176–186, 1997.

[2] W. E. L. Crimson, W. M. Wells III, R. Kikinis, and F. A. Jolesz, “Adaptive segmentation of MRI data,” IEEE Transactions on Medical Imaging, vol. 15, no. 4, pp. 429–442, 1996. [3] N. B. Karayiannis, “A methodology for constructing fuzzy algorithms for learning vector quantization,” IEEE Transactions on Neural Networks, vol. 8, no. 3, pp. 505–518,