165
Table 1. Abbreviation Code for the Various Anatomical Areas in Human Brain Anantomical Area Abbr.Code Superior frontal gyrus ( ) SFG Middle frontal gyrus ( ) MFG Inferior frontal gyrus ( ) IFG Supplementary motor area ( ) SMA Precentral gyrus ( ) PRC Superior temporal gyrus ( ) STG Middle temporal gyrus ( ) MTG Inferior temporal gyrus ( ) ITG Superior parietal gyrus ( ) SPG Inferior parietal gyrus ( ) IPG Supramarginal gyrus ( ) SMG Angular gyrus ( ) AG Precuneus ( ) PCU Postcentral gyrus ( ) POC Superior occipital gyrus ( ) SOG Middle occipital gyrus ( ) MOG Inferior occipital gyrus ( ) IOG Hippocampus ( ) HIP Parahippocampal gyrus ( ) PHIP Amygdala ( ) AMYG Cingulate gyrus ( ) CIN Septal area ( ) SEP Putamen ( ) PUT Globus pallidus ( ) GLO Caudate nucleus ( ) CN Thalamus ( ) THL Hypothalamus ( ) HTHL Insula ( ) INS Corpus callosum ( ) CC Midbrain ( ) MID Medulla ( ) MED Pons ( ) PON Cerebellar cortex ( ) CRBL Vermis ( ) VER 166
Table 2. Correlation of Anatomical Regions and Brodmann s Areas Abbr. Code BA* Function SFG 6,8,9,10,11,32 MFG 6,8,9,10,11,44,45,46,47 IFG 6,11,38,44,45,47 Speech, movement, planning SMA 6 PRC 4,6 STG 21,22,36,38,41,42 Hearing, speech MTG 20,21,22,36,37,38 Form vision ITG 20,37 Form vision SPG 2,5,7 IPG 2,3,40 SMG 2,40,41,42 Reading, speech, movement AG 39 Perception, vision, reading, speech PCU 5,7,23,27,30 Multimodal area for spatial body sence POC 1,2,3,43 SOG 17,18,19 Vision, depth MOG 18,19,39 Vision, depth, color, motion IOG 18,19,37 Vision, depth, color, motion HIP 20,27,28,35,36 Smell, emotion, memory PHIP 27,28,30,35,36 Smell, emotion, memory AMYG 28,34 Emotion, memory CIN 10,11,24,26,29,30,32 Emotion, attention, detection of error SEP PUT 48 Part of involuntary movement, selection GLO of willed movement, regulation, memory CN initiation THL Regulation of emotional behavior HTHL Emotion, function related survival INS 48 Taste, GI tract CC 25 Principal fiber bundle for connection *BA : Brodmann s area 167
이진명 외: 픽셀차분 알고리즘에 의한 대뇌 활성화도의 정성 및 정량 분석 프로그램의 개발 조영상(Fig. 1A)으로 구성된 3차원적인 template(mcconnell Brain Imaging Centre, Montreal Neurological Institute) 에 활성화된 지도를 중첩시켜 이를 그림파일 형태인 활성화 영상(activation image) (Fig. 1B)으로 저장하였다. 이때 T1 강조영상과 활성화 영상은 AC-PC선을 기준으로 상하 -59 mm와 +83 mm의 범위로부터 2 mm 간격으로 총 73개의 단면영상으로 재구성하였다. 이때 각 단면영상은 39,277(181 217)개의 픽셀로 구성되어 있으며, 사용된 영상은 Raw image 형태이고, 각 픽셀은 24 bit의 R, G, B true color로 표현하였다. 일반적으로 24비트 컬러를 표현하기 위해 R, G, B(Red, Green, Blue) 값으로 구분하여 표현할 수 있다. 24 비트로 표현할 수 있는 색상의 범위는 약 1670만(224) 종 류로서, 보통 인간이 자연에서 볼 수 있는 색상을 모두 표현 할 수 있게 때문에 투루컬러라고 부른다. R, G, B컬러는 좌 표계에 따라 각각 1바이트(8비트)씩 할당받게 되고, 따라서 R, G, B의 채도는 각각 28(=256)가지로 나타낼 수 있다. 보 통 인간이 느낄 수 있는 R, G, B 채도값은 보통 120-150정 도 되므로 인간이 분별하기 어려운 처리까지 가능하고, 각 픽 셀들은 R, G, B 각각의 값에 의해 색깔로 표현된다. Table 3 A Fig. 2. A color-cube representing RGB pixel values. D F E G C B Fig. 1. Main protocol for obtaining the anatomical and functional differentiation images. (A) T1-weighted image, (B) activation image, (C) black and white differentiation image, (D) anatomical index image, (E) functional index image, (F) anatomical differentiation image, and (G) functional differentiation image. 168
Table 3. Color Coded RGB Values Corresponding to the Anatomical and Functional Area Anatomical R G B Functional R G B SFG 196 223 155 01 146 039 143 MFG 255 247 153 02 240 030 230 IFG 244 154 193 03 020 245 240 GR 130 202 156 04 255 128 035 OC 146 039 143 05 121 000 000 SCA 068 014 098 06 000 089 082 SMA 000 000 200 07 210 026 110 PRC 000 089 082 08 198 156 109 STG 242 101 034 09 168 252 252 MTG 015 035 235 10 196 223 155 ITG 000 114 188 11 130 202 156 SPG 210 026 110 17 242 109 125 IPG 248 176 172 18 158 011 014 SMG 240 030 230 19 235 010 015 AG 000 084 166 20 000 114 188 PCU 121 000 000 21 015 035 235 POC 020 245 240 22 242 101 034 SOG 102 045 145 23 013 000 076 MOG 235 010 015 24 189 140 191 IOG 141 198 063 25 171 160 000 FUSI 236 000 140 26 099 004 096 LING 158 011 014 27 046 049 146 CU 096 092 168 28 096 057 019 INS 057 181 074 29 000 174 239 HIP 046 049 146 30 130 123 000 PHIP 096 057 019 32 000 000 200 AMYG 255 242 000 34 255 242 000 CIN 189 140 191 35 096 092 168 SEP 082 216 226 36 236 000 140 PUT 033 087 081 37 141 198 063 GLO 000 191 243 38 102 045 145 CN 171 160 000 39 000 084 166 THL 000 165 081 40 248 176 172 HTHL 168 252 252 41 027 020 100 CC 198 156 106 42 068 014 098 PCL 255 128 035 43 000 166 081 RO 013 000 076 44 000 191 243 HES 030 240 020 45 244 154 193 CAL 242 109 125 46 255 247 153 MID 247 148 029 47 33 087 081 PON 163 043 049 MED 080 032 217 CRBL 115 226 120 VER 160 065 013 169
Table 4. Array of RGB Pixel Values Representing a B/W Differentiation Image (Fig. 1C) (1, 1) (2, 1) (101, 1) (102, 1) (103, 1) (180, 1) (181, 1) 0,0,0 (1, 2) (2, 2) (101, 2) (102, 2) (103, 2) (180, 2) (181, 2) 0,0,0 255,255,255 255,255,255 (1, 3) (2, 3) (101, 3) (102, 3) (103, 3) (180, 3) (181, 3) 0,0,0 255,255,255 255,255,255 255,255,255 255,255,255 255,255,255 255,255,255 (1, 216) (2, 216) (101, 216) (102, 216) (103, 216) (180, 216) (181, 216) 255,255,255 255,255,255 255,255,255 (1, 216) (2, 217) (101, 217) (102, 217) (103, 217) (180, 217) (181, 217) 0,0,0 Table 5. Array of RGB Pixel Values Representing a Functional Index Image (Fig. 1E) (1, 1) (2, 1) (101, 1) (102, 1) (103, 1) (180, 1) (181, 1) 0,0,0 (1, 2) (2, 2) (101, 2) (102, 2) (103, 2) (180, 2) (181, 2) 196,223,155 196,223,155 196,223,155 (1, 3) (2, 3) (101, 3) (102, 3) (103, 3) (180, 3) (181, 3) 0,0,0 244,154,193 196,223,155 196,223,155 210,26,110 255,247,153 255,247,153 (1, 216) (2, 216) (101, 216) (102, 216) (103, 216) (180, 216) (181, 216) 235,10,15 235,10,15 158,11,14 (1, 216) (2, 217) (101, 217) (102, 217) (103, 217) (180, 217) (181, 217) 0,0,0 Table 6. Array of RGB Pixel Values for Functional Differentiation Image Obtained by Pixel Differentiation Method from B/W Differentiation and Function Index Images (Fig. 1G) (1, 1) (2, 1) (101, 1) (102, 1) (103, 1) (180, 1) (181, 1) 0,0,0 (1, 2) (2, 2) (101, 2) (102, 2) (103, 2) (180, 2) (181, 2) 59,32,100 59,32,100 59,32,100 (1, 3) (2, 3) (101, 3) (102, 3) (103, 3) (180, 3) (181, 3) 0,0,0 11,101,62 59,32,100 59,32,100 45,229,145 0,8,102 0,8,102 (1, 216) (2, 216) (101, 216) (102, 216) (103, 216) (180, 216) (181, 216) 20,245,240 20,245,240 97,244,241 (1, 216) (2, 217) (101, 217) (102, 217) (103, 217) (180, 217) (181, 217) 0,0,0 170
171
이진명 외: 픽셀차분 알고리즘에 의한 대뇌 활성화도의 정성 및 정량 분석 프로그램의 개발 Fig. 3. An example of measuring the number of activated pixels and brain activities from the anatomical and function differentiation images in a single slice. 172
대한영상의학회지 2004;51:165-177 A B Fig. 4. Quantitative analysis of the number of activated pixels and brain activity in both (A) whole brain area and (B) partially selected brain area, where the slice in red indicates AC-PC line. 173
이진명 외: 픽셀차분 알고리즘에 의한 대뇌 활성화도의 정성 및 정량 분석 프로그램의 개발 Fig. 5. Lateralization indices (%) of cerebrocortical regions based on the anatomical and functional areas, respectively. 174
175
Magn Reson Med 1997;38:296-302 7. Tzourio-Mazoyer N, Landeau B, Papathanassiou D, Crivello F, Etard O, Delcroix N, et al. Automated anatomical labeling of activations in SPM using a macroscopic anatomical parcellation of the MNI MRI single-subject brain. Neuroimage 2002;15:273-289 8. Lancaster JL, Woldorff MG, Parsons LM, Liotti M, Freitas CS, Rainey L, et al. Automated Talairach atlas labels for functional brain mapping. Hum Brain Mapp 2000;10:120-131 9. Kiebel SJ, Poline JB, Friston KJ, Holmes AP, Worsley KJ. Robust smoothness estimation in statistical parametric maps using standardized residuals from the general linear model. Neuroimage 1999;10:756-766 10. Rademacher J, Galaburda AM, Kennedy DN, Filipek PA, Caviness VS. Human cerebral cortex: localization, parcellation, and morphometry with magnetic resonance imaging. J Cogn Neurosci 1992;4:352-374 11. Rajkowska G, Goldman-Rakic PS. Cytoarchitectonic definition of prefrontal areas in the normal human cortex: II. variability in the locations of areas 9 and 46 and relationship to the Talairach coordinates system. Cereb Cortex 1995;5:323-337 12. Rademacher J, Caviness VS Jr, Steinmetz H, Galaburda AM. Topographical variation of the human primary cortices: implications for neuroimaging, brain mapping, and neurobiology. Cereb Cortex 1993;3:313-329 13. Caviness VS Jr, Meyer J, Makris N, Kennedy DN. Specified method with estimate of reliability. J. Cogn. MRI-based topographic parcellation of human neocortex: An anatomically. Neuroscience 1996;8:566-587 14. Wu DC, Tsai WH. A steganographic method for images by pixelvalue differencing. Pattern Recognit Lett 2003;24:1613-1626 15. Philipp I, Rath T. Improving plant discrimination in image processing by use of different color space transformations. Comput 1. Ogawa S, Lee TM, Nayak AS, Glynn P. Oxygenation-sensitive contrast in magnetic resonance image of rodent brain at high magnetic Electron Agric 2002;35:1-15 field. Magn Reson Med 1990;14:68-78 16. Hwang KF, Chang CC. A fast pixel mapping algorithm using principal component analysis. Pattern Recognit Lett 2002;23:1747-1753 2. Ogawa S, Lee TM, Kay AR, Tank DW. Brain magnetic resonance imaging with contrast dependent on blood oxygenation. Proc Natl 17. Yue ZQ, Chen S, Tham LG. Finite element modeling of geomaterials using digital image processing. Comp Geotech 2003;30:375-397 Acad Sci USA 1990;87:9868-9872 3. Duyn JH, Moonen CT, Van Yperen GH, de Boer RW, Luyten PR. 18. Zhao W, Busto R, Truettner J, Ginsberg MD. Simultaneous measurement of cerebral blood flow and mrna signals: pixel-based in- Inflow versus deoxyhemoglobin effects in BOLD functional MRI using gradient echoes at 1.5T. NMR Biomed 1994;7:83-88 ter-modality correlational analysis. J Neurosci Methods 2001;108: 4. Duong TQ, Silva AC, Lee SP, Kim SG. Functional MRI of calciumdependent synaptic activity: cross correlation with CBF and BOLD 161-170 19. Friedlinger M, Schro der J, Schad LR. Ultra-fast automated brain measurements. Magn Reson Med 2000;43:383-392 volumetry based on bispectral MR imaging data. Comput Med 5. Lee SP, Silva AC, Ugurbil K, Kim SG. Diffusion-weighted spinecho fmri at 9.4 T: microvascular/tissue contribution to BOLD sig- Imaging Graph 1999;23:331-337 20. Yule B, Roberts S, Marshall JE. Quantitative spore color measurement using image analysis. Org Geochem 1998;28:139-1490 nal changes. Magn Reson Med 1999;42:919-928 6. Gati JS, Menon RS, Ugurbil K, Rutt BK. Experimental determination of the BOLD field strength dependence in vessels and tissue. 176
Qualitative and Quantitative Measurement of Human Brain Activity Using Pixel Subtraction Algorithm 1 Jin-Myoung Lee, Gwang-Woo Jeong, Ph.D. 1,2, Hyung-Joong Kim, Ph.D. 2, Seong-Hoon Cho, Heoung-Keun Kang, M.D. 2, Jeong-Jin Seo, M.D. 2, Seung-Jin Park, Ph.D. 3 1 Interdisciplinary Program of Biomedical Engineering, Chonnam National University Graduate School 2 Department of Radiology, Chonnam National University Medical School 3 Department of Biomedical Engineering, Chonnam National University, Hospital Purpose: To develop an automated quantification program, which is called FALBA (Functional & Anatomical Labeling of Brain Activation), and to provide information on the brain centers, brain activity (%) and hemispheric lateralization index on the basis of a brain activation map obtained from functional MR imaging. Materials and Methods: The 3-dimensional activation MR images were processed by a statistical parametric mapping program (SPM99, The Wellcome Department of Cognitive Neurology, University College London, UK) and MRIcro software (www.mricro.com). The 3-dimensional images were first converted into 2-dimensional sectional images, and then overlapped with the corresponding T1-weighted images. Then, the image dataset was extended to 59 mm to 83 mm with a 2 mm slice-gap, giving 73 axial images. By using a pixel subtraction method, the differences in the R, G, B values between the T1-weighted images and the activation images were extracted, in order to produce black & white (B/W) differentiation images, in which each pixel is represented by 24-bit R, G, B true colors. Subsequently, another pixel differentiation method was applied to two template images, namely one functional and one anatomical index image, in order to generate functional and anatomical differentiation images containing regional brain activation information based on the Brodmann's and anatomical areas, respectively. In addition, the regional brain lateralization indices were automatically determined, in order to evaluate the hemispheric predominance, with the positive (+) and negative ( ) indices showing left and right predominance, respectively. Results: The manual counting method currently used is time consuming and has limited accuracy and reliability in the case of the activated cerebrocortical regions. The FALBA program we developed was 240 times faster than the manual counting method: 10 hours for manual accounting and 2.5 minutes for the FALBA program using a Pentium IV processor. Compared with the FALBA program, the manual quantification method showed an average error of 0.334 0.007 (%). Thus, the manual counting method gave less accurate quantitative information on brain activation than the FALBA program. Conclusion: The FALBA program is capable of providing accurate quantitative results, including the identification of the brain activation region and lateralization index with respect to the functional and anatomical areas. Also, the processing time was dramatically shortened in comparison with the manual counting method. Index words : Brain Brain, function Brain, MR Address reprint requests to : Gwang-Woo Jeong, Ph.D., Department of Radiology, Chonnam National University Medical School 8 Hack-dong, Dong-gu, Kwang-ju 501-757, Korea. Tel. 82-62-220-5881 Fax. 82-62-226-4380 E-mail: gwjeong@chonnam.ac.kr 177