**3. Experimental results**

4 Advances in Brain Imaging

In the equation, *BB B* 1 , 2 and 3 indicates the mean of the selected 3×3 range, calculated as

9

<sup>1</sup> 1 1 9

*B B*

 

 

each pixel vector and its eight neighboring points was calculated as follows:

1,2,...,8

The standard deviation was further normalized by [0, 1] as follows:

pixel vector and its eight neighbors was calculated as follows:

*i*

the following equation:

between two regions.

**2.2 Region growing** 

*i*

The overall standard deviation is calculated as follows:

*B B*

*B B*

 

*i*

*i*

max indicated the maximum standard deviation in the image. The similarity between a

Furthermore, according to the second condition, the relative Euclidean distance between

222

*iii*

222 ( 1 1) ( 2 2) ( 3 3) 123

*BBB*

According to the experiment by Shih & Cheng [11], the efficacy of employing relative Euclidean distance is better than that of using normal Euclidean distance. Therefore, the maximum distance between each pixel and its eight neighboring points was calculated by

> 8 max <sup>1</sup>

The two aforementioned conditions have to be satisfied in order for a pixel vector to be selected as a seed point. The first condition aims at examining whether or not there is considerably high similarity between a seed pixel vector and its neighbors whereas the second condition focuses on ensuring that a seed pixel vector is not in the boundary

The traditional method of region growing can not successfully classify brain tissues. Consequently, this paper modified the principle of region growing and recorded all the

max( )*<sup>i</sup> <sup>i</sup> d d*

*H* 1 

*BB BB BB <sup>d</sup>*

*i*

1 9

*i*

*i*

(2)

*BBB* 123 (3)

*<sup>N</sup>* / max (4)

*<sup>N</sup>* (5)

(7)

(6)

*i*

1 9

1

follows:

The real MR images were used for performance evaluation. They were acquired from ten patients with normal physiology. One example is shown in Fig. 2(a)-(e) with the same parameter values in Table 1. Band 1 is the PD-weighted spectral image acquired by the pulse sequence TR/TE = 2500ms/25ms. Bands 2, 3 and 4 are T2-weighted spectral images were acquired by the pulse sequences TR/TE = 2500ms/50ms, TR/TE = 2500ms/75ms and TR/TE =2500ms/100ms respectively. Band 5 is the T1-weighted spectral image acquired by the pulse sequence TR/TE = 500ms/11.9ms. The tissues surrounding the brain such as bone, fat and skin, were semiautomatically extracted using interactive thresholding and masking. The slice thickness of all the MR images are 6mm and axial section were taken from GE MR 1.5T Scanner.

In this experiment, there was one type of real brain MR images. In order to evaluate the performance of the UVSRG, the widely used c-means method (also known as k-means) is used for comparative analysis. The reason to select the c-means method is because it is a spatial-based pattern classification technique. In order to make a fair comparison, the implemented c-means method always designates the desired target signature d as one of its class means with d fixed during iterations.

In order to enhance classification of these MR images, the interfering effects resulting from tissue variability and characterization must be eliminated. However, to identify the sources that cause such interference is nearly impossible unless prior information is provided. On the other hand, in many MRI applications, the three cerebral tissues, GM, WM and CSF are of

Automatic Vector Seeded Region Growing for Parenchyma Classification in Brain MRI 7

*N ND(d) NF(d) RD(d) RF(d) RD RF*

*N ND(d) NF(d) RD(d) RF(d) RD RF*

(a) (b) (c)

(d) (e)

Fig. 2. Real brain MR images. (a) TR1/TE1=2500ms/25ms (b) TR2/TE2=2500ms/50ms (c) TR3/TE3=2500ms/75ms (d) TR4/TE4=2500ms/100ms (e) TR5/TE5=500ms/11.9ms

WM 8745 8290 352 0.9479 0.0285 0.9616 0.0280

WM 8745 8737 4 0.9990 0.0003 0.9994 0.0004

GM 9040 9036 8 0.9995 0.0006

GM 9040 8688 455 0.9610 0.0378

CSF 3282 3282 0 1 0

CSF 3282 3282 0 1 0

Table 4. Detection results with SNR =10 db

Table 5. Detection results with SNR = 5 db

major interest where their knowledge can be generally obtained directly from the images. A zero-mean Gaussian noise was added to the phantom images so as to achieve various signalto-noise ratios (SNR) ranging from 5db to 20db. Table 1 tabulates the values of the parameters used by the MRI pulse sequence and the gray level values of the tissues of each band used as phantom in the experiments and Tables 2-5 tabulate the results for SNR = 20db, 15db, 10 and 5db respectively. In our experiments, the spectral signatures of GM, WM and CSF used for the UVSRG were extracted directly from the MR images. Fig. 3(a)-(c) show the classification results of the UVSRG using five images in Fig. 1(a)-(e). For comparison, we also applied the cmeans method to Fig. 2(a)-(e) to produce Fig. 4(a)-(c) where the classification maps of GM, WM and CSF are labeled by (a), (b) and (c) respectively. Compared to Fig. 3(a)-(c), the UVSRG performed significantly better than did the c-means method. All the experimental results presented here were verified by experienced radiologists.


Table 1. Gray level values used for the five bands of the test phantom


Table 2. Detection results with SNR = 20 db


Table 3. Detection results with SNR = 15 db


Table 4. Detection results with SNR =10 db

major interest where their knowledge can be generally obtained directly from the images. A zero-mean Gaussian noise was added to the phantom images so as to achieve various signalto-noise ratios (SNR) ranging from 5db to 20db. Table 1 tabulates the values of the parameters used by the MRI pulse sequence and the gray level values of the tissues of each band used as phantom in the experiments and Tables 2-5 tabulate the results for SNR = 20db, 15db, 10 and 5db respectively. In our experiments, the spectral signatures of GM, WM and CSF used for the UVSRG were extracted directly from the MR images. Fig. 3(a)-(c) show the classification results of the UVSRG using five images in Fig. 1(a)-(e). For comparison, we also applied the cmeans method to Fig. 2(a)-(e) to produce Fig. 4(a)-(c) where the classification maps of GM, WM and CSF are labeled by (a), (b) and (c) respectively. Compared to Fig. 3(a)-(c), the UVSRG performed significantly better than did the c-means method. All the experimental results

Band # MRI Parameter BKG GM WM CSF

Band 1 TR/TE=2500ms/25ms 3 207 188 182

Band 2 TR/TE=2500ms/50ms 3 219 180 253

Band 3 TR/TE=2500ms/75ms 3 150 124 232

Band 4 TR/TE=2500ms/100ms 3 105 94 220

Band 5 TR/TE=500ms/11.9ms 3 95 103 42

WM 8745 8745 0 1 0 1 0

WM 8745 8745 0 1 0 1 0

*N ND(d) NF(d) RD(d) RF(d) RD RF*

*N ND(d) NF(d) RD(d) RF(d) RD RF*

Table 1. Gray level values used for the five bands of the test phantom

GM 9040 9040 0 1 0

CSF 3282 3282 0 1 0

GM 9040 9040 0 1 0

CSF 3282 3282 0 1 0

Table 2. Detection results with SNR = 20 db

Table 3. Detection results with SNR = 15 db

presented here were verified by experienced radiologists.


Table 5. Detection results with SNR = 5 db

Fig. 2. Real brain MR images. (a) TR1/TE1=2500ms/25ms (b) TR2/TE2=2500ms/50ms (c) TR3/TE3=2500ms/75ms (d) TR4/TE4=2500ms/100ms (e) TR5/TE5=500ms/11.9ms

Automatic Vector Seeded Region Growing for Parenchyma Classification in Brain MRI 9

[3] Saif D. Salman & Ahmed A. Bahrani, "Segmentation of tumor tissue in gray medical

[4] J. R. Jimenez-Alaniz, V. Medina-Banuelos, O. Yanez-Suarez, "Data-driven brain MRI

[5] T. Song, M. M. Jamshidi, R. R. Lee, M. Huang, "A Modified Probabilistic Neural

[6] J. Tohka, E. Krestyannikov, I. D. Dinov, A. M. Graham, D. W. Shattuck, U.

[7] S. Duchesne, A. Caroli, C. Geroldi, C. Barillot, G. B. Frisoni, D. L. Collins, "MRI-Based

[8] J. J. Corso, E. Sharon, S. Dube, S. El-Saden, U. Sinha, A. Yuille, "Efficient Multilevel Brain

[9] A. Mayer, H. Greenspan, "An Adaptive Mean-Shift Framework for MRI Brain Segmentation", IEEE Transactions on Medical Imaging 28 (8) (2009) 1238-1250. [10] H. Gudbjartsson, S. Patz, "The Rician distribution of noisy MRI data," Magn. Reson.

[11] C. M. Wang, C. C. C. Chen, Y. N. Chung, S. C. Yang, P. C. Chung, C. W. Yang, C. I.

[13] R. Adams, L. Bischof, "Seeded region growing," IEEE Trans. on Pattern Analysis and

[14] T. Pavlidis, Y.T. Liow, "Integrating region growing and edge detection," IEEE Trans. on Pattern Analysis and Machine Intelligence, vol.12, no.3, pp. 225–233, 1990. [15] N.R. Pal, S.K. Pal, "A review on image segmentation techniques," Pattern Recognition,

[16] C. Chu, J.K. Aggarwal, "The integration of image segmentation maps using region and

[17] A. Tremeau, N. Bolel, "A region growing and merging algorithm to color segmentation," Pattern Recognition, vol.30, no.7, pp.1191–1203, 1997. [18] S.A. Hojjatoleslami, J. Kittler, "Region growing: a new approach," IEEE Trans. on Image

edge information," IEEE Transactions on Pattern Analysis and Machine

IEEE Transactions on Medical Imaging 27 (4) (2008) 509-520.

Transactions on Medical Imaging 27 (5) (2008) 629-640.

Trans. on Medical Imaging, vol.25, no.1, pp.74–83, 2006.

Machine Intelligence, vol.16, no.6, pp.641–647, 1994.

Intelligence, vol.15, no.2, pp.1241–1252, 1993.

Processing, vol.7, no.7, pp.1079–1084, 1998.

Med. vol.34, no.6, pp.910–914, 1995.

vol.26, no.9, pp.1277–1294, 1993.

Transactions on Medical Imaging 25 (1) (2006) 74–83.

on Neural Networks 18 (5) (2007) 1424-1432.

127, 2010

(5) (2007) 696-711.

images using watershed transformation method", IJACT, Vol. 2, No. 4, pp. 123 ~

segmentation supported on edge confidence and a priori tissue information", IEEE

Network for Partial Volume Segmentation in Brain MR Image", IEEE Transactions

Ruotsalainen, A. W. Toga, "Genetic Algorithms for Finite Mixture Model Based Voxel Classification in Neuroimaging", IEEE Transactions on Medical Imaging 26

Automated Computer Classification of Probable AD Versus Normal Controls",

Tumor Segmentation With Integrated Bayesian Model Classification", IEEE

Chang, "Detection of Spectral Signatures in Multispectral MR Images for Classification," IEEE Trans. on Medical Imaging, vol.22, no.1, pp.50-61, 2003. [12] J. R. Jimenez-Alaniz, V. Medina-Banuelos, O. Yanez-Suarez, "Data-driven brain MRI

segmentation supported on edge confidence and a priori tissue information," IEEE

Fig. 3. Real Classification result of brain MR images by UVSRG. (a)GM (b)WM (c)CSF

Fig. 4. Real Classification result of brain MR images by C-means. (a)GM (b)WM (c)CSF
