ѧ
JOURNAL OF SOFTWARE
1999 6 10 Vol.10 No.6 1999



ͼ׺ģĴŹͼָע
̷ Ÿ ִ  λ άѩ
ժҪTT AtlasзḻĽṹϢ,һԶָMRI(magnetic resonance image)ͼķ.ַɷΪ.,MRIͼTT Atlas׼,ͨͼҽѧͼ׵ƥ,ͼнṹϢ֪ʶ,ͿԶͼķָע.Ȼ,ԤָģMRIͼģָ,Ӷ߷ָľ.ΪԶؽԤģеĽṹϢڷָ,»һ״ӵFCM㷨.ƥʱҪֹһЩ֮,÷Զ.
ؼʡMRI(magnetic resonance image),Զͼָ,ͼƥ,ͼ,ģCֵ,״,ӻݼ.
ͼšTP391

Segmentation and Classification of MRI Based on FCM Clustering
Algorithm and Computerized Brain Atlas
TAN Ou JIA Chun-guang L Xu-dong DUAN Hui-long LWei-xue
(Department of Life Science and Biomedical Engineering Zhejiang University Hangzhou 310027)
Abstract An automatic segmentation and classification method of magnetic resonance image of human brain based on a computerized brain atlas-TT atlas is introduced in this paper. The algorithm could be divided into two steps. Firstly, the object brain dataset is registered with the TT atlas by a 3D image matching method. With the help of anatomy structure information of the computerized atlas, the brain image is segmented and classified coarsely. Secondly, the segmentation result is taken as the input of a fuzzy cluster segmentation. In order to use the anatomic structure information directly, a new fuzzy cluster method with a shape factor is introduced, the method is mostly and automatically except selection of some intrinsic point pairs for global linear matching method in the first step.
Key wordsMRI(magnetic resonance image), automatic image segmentation, image matching, brain atlas, FCM clustering, shape factor, visible human dataset.
MRI(magnetic resonance image)ķָעָ,MRͼͬһ֯ͼԪȺ,ΪͬһࣨͼԪǶά,;Ҳά,أ.ĿΪͬ֯͵״ߴ,Ȼʾάռ,ָ,,ļ⡢ԲٲϡƵļƻ.
ڹȥʮ,ԶϲͼԶָͱע,һĳɹ13.ָһɷΪ:ڱԵĺͻ.ȻԵⷽԲյı߽,ڱԵӶŽ,ڱ߽粻̫Եϴ,ڵעʶ.ʶֵ෨һԲֿ-ϲԼЩϵ.,ҽѧͼϸӵĶ,ҪͼȫԶطָе֯ṹ,Ƿǳѵ.ͼƥǷࣩͼһֵԻһԵĴ.ͨҽѧͼҽѧͼ׵ƥ䣬԰Ѳͬģ̬ͼͳһĳ֪ı׼ϵͳռ滯,Ӷȷͼݼ֮ļα任ϵ,ҿʵҽѧͼָµҽѧͼָͷ.ҽѧͼҽѧͼ֮ƥ,÷ǸԵƥ䷽.׼ȷƥ԰ҽѧͼ׵﷨ĺ/Ϣӳ䵽ͬ˸ͼȥ.ҽѧͼͼ֮ƥ㷨Ըƥ˹ԤĶٷΪֹ׼ָĶ׼ȫԶĶ׼4.
͵ֻҽѧͼзTalairachTournouxTalairach-Tournouxͼ(TT atlas)5,¹ѧHhneVoxelMan6,ҽѧͼ(NLM)ʵֵֻVisible Human Project7ݵ.
FCM㷨һЧݾ෽.ͨŻһģĿ꺯ʵ־,һҪͨ.㷨صǸÿݵһŶ,ݵԸ̶,ӲΪÿֻĳһض.
FCM㷨ͼָѵõ㷺Ӧ.ϷݴѧMatthew C.Clark˽ָFCM㷨֪ʶģ,ʵMRIͼĶάԶָ8.ǵģֻ3Ƭ,ָҪڻʡʵȽϴĽṹ,ҷָͱעǷֱе.
һԶָMRIͼķ.ַɷΪ.,MRIͼTT Atlas׼,ͨͼҽѧͼ׵ƥ,԰ͼд洢Ϣֱӳ䵽MRI.Ȼ,ԤָģFCMģָ.ĻһּȨFCM㷨,ģϢֱӷӳFCM㷨ĵ,ķٶ.ĵ1ڷͼƥͼȨFCMָ㷨ԭ˾ķͲѡ.2ں͵3ڶʵ˱ȽϺͷ.ǽ.
1  
1.1 ƥ 
ͼƥ䷽һֱӵӦþʵҽѧͼԶָ.ҽѧͼϸӵĶ,ҪͼȫԶطָе֯ṹ,ǲܵ.ͨͼҽѧͼ׵ƥ,԰ͼд洢Ϣֱӳ䵽ָͼ,ͺñǽר֪ʶָķָ,̿ʵȫԶ.Ϣ,ϢԹϵԼϢ,Ҳͨƥӳ䵽ͼȥ.
ǲTTͼΪƥԴͼ.ΪṩһĲοϵ,мΪḻĿռʽṹϢ.,Ȼͼ׺κʵʲ˵ͼûֱӵĹϵ,ҽѧѧżΪ㷺Ӧ.
ƥ䷽ϿɷΪԺͷ.ǵҪ˵,һϺõάֶƥ͹.ĺôƥʱֹעԼ.
ƥĲƽơתšȣͨȷ,:
1 ͼƥ:ûѡһϵеĲοƽΪͼ,Ȼϵͳƥ任.ҵи,Ȼҵǰϵ㣨ACͺϵ㣨PC,󶨳ԵĶ,иͨAC㴹ֱAC-PCƤ㶥Ľ.Щο,ǾͿTTͼ׵ϵƥĲ.ϵͳѱ֤ԽṹĶλ.Ķ:ԭAC,1AC-PCȷ,2ķиķߵķȷ,3ᴹֱǰᲢҷ.Ϊȷˮƽų߶,ǻҪһЩֵ.
2 Ӿƥ:ûͨȽĿͼͼ׵ĽһЩα任Ĳ.ڴĳЩͼʱ,ֱˮƽűȶֵɿ.
ʵӦ,ַʹ.һͼı任,ٸݽһЩӾƥ.
1.2 ״ӵģָ
ڶͼɫͼMRI3Ƶʷɵͼ񣬻ݶȡȵͼ񣩽зָ,Բþ෽.FCM㷨ͼľ࣬˿Եõָͼ֮,Ըصжϸĳһ̶.ԶͼطǺõ.磬ĳԸԶС1ʱ,䴦ڸķֽ紦,ĳĻ.
ͨͼƥĽ,ǿɵõһԤָģ壬ӳиṹĿλá״.Ϊ˽Щ״Լָ.ǽĿͼصĿռΪһ.,ڸ֯ṹ״,λ״ʾ.,㵽ĵľ벻ŷ¾ʾ.,״.״,ǿԼռĳһ㵽ĳĿռ,ռľȻŷ¾.Ȼ־мȨ,͵õͨģʹõľ.
ν״,򵥵˵,һļʾĳṹ,ĳ㵽ýṹľΪõ㵽ĵ㼯̾.ռе㵽ýṹľľ󹹳ɸýṹ״.Ӧ:1 ״ĵԽ,ԽС;2 ״Եĵ㵽ĵľ.,㼯ʵǺѼõ.ѧ̬еĸʴͲ,Աܿ㼯,ֱӼ㵽ĳľ.
,ÿһṹMһֵģ.di,j,kʾӵ(i,j,k)Mľ,ļ㷽.
(1) ʼ,l=0,M(l)=M,
(1)

(2) 
(2)

(3) 
M(l+1)=erosion(M(l)).(3)

(4) M(l+1)=M(l),(5),l=l+1,ص(2).
(5) di,j,k(0)=di,j,k(l), l=0,M(l)=M.
(6) Ͳ
M(l+1)=dilation(M(l)).(4)

(7) 
(5)

(8) l(t*L)Mk(l+1)=Mk(l),㷨,l=l+1,ص(6).
ģľ,d(i,j,k)ͼռһ㣨i,jkľ,dΪ(4)õľ.
(6)

ʽa3Ϊϵ,cΪܵĽṹ.
˾ļ㷽,ǿɷմͳFCM㷨Ĳģָ.ͼ1Ϊṹ(Ƥʡ״ˡԡ)״,ԽʾԽ,ǵļ໥Լ.

 
 
ͼ1ĳһ4ֽṹ״
1.3 㷨
1 TTͼûͷƤ­ǡȲ,ģָǰ,ҪMRIͼȥЩ.Matthew C.Clarkף8иһַ,Էȥ­ǵ֯.˼:T1Ȩͼֵ.­ǵT1ֵС,׽Էֿ,Ȼѧ̬ѧķõԵĶֵģ.
2 TTͼ׺ԵMRIͼάƥ.ͨ,иACҵ,Ƭ̫ʱ5mm,PCʱҲ,ʱ,ɸԹPC.²ɵõʸ״Ƭ.иϿȷԵĶ.Ȼ,ǿԸЩο,ͼÿһTTϵе.ٽһЩ,磬ʹƤغTTͼ׵Ƥ൱,ʹͼTTͼ׸õƥ䣨λϣ.ɴ,ƶMRIÿһάϲͼ񣬿ɵһ֯ṹϢԤָģ.
3 Ԥָģ,յ1.2е㷨״Dik.
4 ״ӵFCM,ھ,ĳ㵽ĳĿռ뱣ֲ.
dik=a1Dik+a2dik,(7)

dikʽ(7).ʱ,vk(T1,T2,PD)ɵR3ռе;a1a2Ǳϵ.ĵ̺ͱ׼FCMͬ.
5 ͨ,ָMRIͼ.
6 ظ(3)(5).
1.4 ѡ
,漰:һFCMĵ;һǷָ״ӵĴ.ڵ1,Ȼ󷨺ͱ׼ͬ,ڿռڵʱֵֶ,,õһ.,ڳʼȾɷָģõ,һŻĽ,,ҲǱȽϿ.ڵ2,Խ,ԤָģԽ,ָͼԵԾԽǿ.޶ʱ,ͳһFCMָ.ǿģ,Ҫ.֮,ӵ.MRIͼָ,Ӧ,һȡ24.
2 ָ
NLMVisible Human DatasetеMRIͼ㷨֤.Ƕͼƥ,ƥ䷽ҵͼ񼯵ϵTTϵı任,ȻĳһӦTT֯,ͼ2aΪTTͼеһƬͼԭͼΪɫ,ͼ2bΪƥҼ򻯺õԤָģ,ͼ2(c)ȥ­ǵ֯MRIͼԭͼΪT1,T2,PD)Ĳɫͼ,ͼ2(d)ģģָĽ,ͼ2e״ӵģָ,ε,ͼ2(f)״ӵģָ,Ĵε.ͼ2b˰˷õľͷ90cmMRIƬԤָģ.Ȼ,ͼһЩṹ˺ϲͼ.õģƤʡʡβ״ˡ״ˡԡҡȽṹ.ͼ2e˷ָĽ.FCM㷨ģΪ2,ģĵΪ2,FCMеΪ100,ֹΪţ10-6.ڽṹϢֱӺ,ǽͬòͬɫʾ.ӷָ,ṹнΪı߽,ҺָֹЧ.״ʧĽ,״ӵģָܽԽṹϢʵͼ,ﵽԶָͱעĿ.


(a) (b) (c)


(d) (e) (f)
ͼ2 ָ
ʵʷָ,һЩṹûʲô,ֻڽ,ͨ򵥵ģҲ޷ֿ:Ͱ;β״˺Ͷ״,״ӵģָܵõ,߽ԲΪ,Ҳûʵüֵ.Ľṹ,ǲöηָķ,Ƚϲ,ΪͬһƥָȻģֽṹĿռϵӳ䵽ָĽ.Ȼ,ҪԷָͼз,MRIͼƵ,ǿͳ,ӦõеMRIͼıע.
3 
ͼ3ΪԵļ,кΪʱĵ,ΪĿ꺯ֵ.ʵΪ1μĿ꺯,Ϊ2,ԲȦΪ3,Ϊ4.Ϊ,ö.ģÿһ,ǿɵõһĿ꺯,ͼ3ģ4õ4.ͼǿԿ,ÿεǱȽϿ,رģԺ,ɷָõģȾĿ꺯ֵѾС,ٶҲȳʼģ.3Ρ4εĿ꺯ֵٶȺ͵2μͬ,ͼϺѽ߷뿪.ΪΪģѾʵʽ,ٴξĿ꺯仯.ʱڿռϸĵ㼯ָ仯ڽ,ڿռ,ֵ,õķָ仯.


ͼ3 Ŀ꺯仯ͼ
4  
˽ͼƥģָһԶMRIͼķָע.,״ӵģָܽԽṹϢʵͼ,ʵMRIͼԶָͱע,رвĽṹ,ԡǡ԰ȵķָע.,ڻڽͼͬ,Ҳ,,ֻҵϲԵͱע,֮ķֻֽģ,ʶЧͬ.һĹȡͼ,,ɽһŻָĽͷָ.ķҲϲͼķָע,ԵĲɫƬͼCTͼȵԶָע.
߼顡̷Ÿ,1973,ʿ,ҪоΪҽѧͼ,ʵ.
ִ,1969,ʿ,ҪоΪźżҽѧͼ,ͼѧ.
,1973,ʿ,ҪоΪҽѧͼ,Զҽ.
λ,1963,ʿ,,ҪоΪӾ,źżҽѧͼ,Զҽ.
άѩ,1930,,ʿʦ,ҪоΪҽѧϢ,ͼؽ,潨ģ.
ͨѶϵ:̷Ÿ 310027,㽭ѧѧҽѧѧϵ590#
ߵλ㽭ѧѧҽѧѧϵ  310027
ο
1 Fu K, Mui J. Survey on image segmentation. Pattern Recognition, 1981,13(1):316
2 Li C, Goldgof D, Hall L. Automatic segmentation and tissue labeling of MR brain images. IEEE Transactions on Medical Imaging, 1993,12(12):740750
3 Liang Zheng-rong. Tissue classification and segmentation of MR images. IEEE Engineering in Medicine and Biology, 1993,3(3):8184
4 Petra A, van den Elsen et al. Medical image matching-a review with classification. IEEE Engineering in Medicine and Biology, 1993,3(3):2630
5 Talairach J, Tournoux P. Co-planar Stereotaxic Atlas of the Human Brain 3-dimensional Proportional System: An Approach to Cerebral Imaging. New York: Georg Thieme Verlag, Stuttgart, 1988
6 Hohne K H, Pflesser B et al. A new representation of knowledge concerning human anatomy and function. Nature Medicine, 1995,1(6):506511
7 Ackerman M J. The visible human project. National Library of Medicine, World Wide Web address: http:www.nlm.nih.gov/, 1995
8 Clark M C, Hall L O. MRI segmentation using fuzzy clustering techniques: integrating knowledge. World Wide Web address: http: www.csee.usf.edu/, 1995
1998-03-23յԭ,1998-06-29յ޸ĸ
