104 김덕임, 이우영, 한승호 자의뼈가백인여자머리뼈처럼보이기도하며그반대의경우로추정되기도한다 [2,4-8]. 2002년부터 2007년까지국립과학수사연구소에의뢰된 290건의머리뼈 (skull) 중아래턱뼈를포함한머리뼈는 52%, 온전한형태로발견된머리뼈 (cranium) 와

Transcription

1 대한체질인류학회지제 28 권제 2 호 Korean J Phys Anthropol Vol. 28, No. 2 (2015) pp. 103~118 Original Article 한국인머리뼈 3 차원영상을이용한성별판별 : 판별분석을통한계측학적연구 김덕임 1, 이우영 2, 한승호 3 1 계명대학교간호대학간호학과, 2 가톨릭대학교의과대학해부학교실 가톨릭응용해부연구소 3 중앙대학교의과대학해부학교실 (2015 년 4 월 28 일접수, 2015 년 6 월 9 일수정접수, 2015 년 6 월 15 일게재승인, Published Online 30 June 2015) 간추림 : 신원확인이필요한백골화시신이발견된경우성별판별은가장먼저시행되는과정으로머리뼈는성별판별에높은정확도를보이는뼈들중하나이다. 한국인머리뼈를계측하여계측값으로성별을비교한연구는있었으나판별방정식을제안한연구는없었다. 이연구의목적은한국인 3차원머리뼈영상을이용하여누구나사용하기쉽고정확한판별방정식을만들어제공하는것에있다. 102구시신의 CT 영상을컴퓨터프로그램을이용하여 3차원재구성하고계측템플릿을작성하여총 44개의항목을계측하였고항목들의급내상관계수, 효과크기, 곡선아래영역값을기준으로정확도와신뢰도가높은항목을분류하여판별분석하였다. 정확도가가장높은방정식을사용한경우남자가 82.7%, 여자는 82.2% 정확하게판별되었으며방정식을구성하는항목은얼굴너비 (BiZ D), 머리뼈바닥길이 (Cb L), 귓바퀴사이너비 (BiA B), 위얼굴너비 (UpF B), 이마직선길이 (FC), 큰구멍너비 (FMa B), 그리고오른쪽꼭지돌기길이 (Ma L, right) 이었다. 한국인 3차원재구성머리뼈영상을이용한성별판별분석의교차유효값은최소 76.5% 에서최대 86.7% 이었다. 다른인구집단과의머리뼈계측값을비교해보면일치하는항목대부분이통계학적으로유의한차이가있었으며남자의경우유럽인보다크기가작았고여자의경우동양인보다크기가컸다. 이연구의결과는한국인머리뼈를이용한성별판별에도움이될것이라생각한다. 찾아보기낱말 : 3 차원영상, 머리뼈, 판별분석, 계측연구, 한국인 서 론 신원을확인할수없는백골화시신이발견된경우신원 을밝히기위해성별 (sex), 사망당시나이 (age at death), 키 (stature), 인구집단 (ethnic origin) 을추정한다. 그중성별판 이논문은 2011년정부 ( 교육과학기술부 ) 의재원으로한국연구재단의지원을받아수행된연구임. [NRF E00003] 저자 ( 들 ) 는 의학논문출판윤리가이드라인 을준수합니다. 저자 ( 들 ) 는이연구와관련하여이해관계가없음을밝힙니다. 교신저자 : 한승호 ( 중앙대학교의과대학해부학교실 ) 전자우편 : monami@cau.ac.kr 별 (sex determination) 은신원확인을위한과정중가장먼저시행되는중요한단계로성별판별의정확도와신뢰도는뼈가온전한지, 성별의특징이얼마나뚜렷하게나타나는인구집단 (population group) 인지등에의해결정된다 [1-3]. 성별판별에사용되는뼈는머리뼈 (cranial bone) 와머리뼈이후뼈대 (postcranial skeleton) 로분류되며그중가장정확도가높은것은머리뼈 (skull) 와골반뼈 (pelvic bone) 로알려져있다 [3,4]. 머리뼈는형태학적특성을가장잘나타내며환경적요인에영향을가장적게받는부위로성별을판별하는데유용한뼈이지만, 종종인구집단간의차이로인해동양인남 c 2015 Korean Association of Physical Anthropologists This is an Open Access article distributed under the terms of the Creative Commons Attribution Non-Commercial License ( licenses/by-nc/3.0) which permits unrestricted non-commercial use, distribution, and reproduction in any medium, provided the original work is properly cited. ISSN X (Online) ISSN X (Print)

2 104 김덕임, 이우영, 한승호 자의뼈가백인여자머리뼈처럼보이기도하며그반대의경우로추정되기도한다 [2,4-8]. 2002년부터 2007년까지국립과학수사연구소에의뢰된 290건의머리뼈 (skull) 중아래턱뼈를포함한머리뼈는 52%, 온전한형태로발견된머리뼈 (cranium) 와아래턱뼈는각각 60% 정도였고나머지 40% 는발견되지않았거나파손된상태로발견되었다 [9]. 이와같이머리뼈의경우발굴현장이나범죄현장에서는부분적혹은전체적으로파손된상태로발견되는경우도많아실제머리뼈를이용한신원확인에는어려움이있다. 컴퓨터단층촬영이나방사선촬영으로얻은영상들은객관적이고사정할물적증거를영구적으로보관할수있으며사전 (antemortem), 사후 (postmortem) 분석을비교할수있는자료를제공하는장점이있다. 특히컴퓨터단층촬영의경우유용성이증가되고있으며, 많은사람들이일생동안촬영하는경우가한번이상있어많은경우사전자료가확보되어있으며또한점차적으로기존의방사선영상대신컴퓨터단층촬영영상으로교체되고있는추세이다 [3,10]. 컴퓨터단층촬영영상을이용하여 3차원으로재구성한경우실제뼈를계측한것과같은값을얻을수있으며축을따라쉽게이미지를돌려볼수있는장점들이있어 3차원영상은연구와교육에폭넓게사용되고있다 [11]. 이연구의목적은개발된컴퓨터프로그램용템플릿을이용하여한국인머리뼈 3차원재구성영상에알맞은성별판별방정식을만들고, 다른인구집단의계측값과판별방정식을비교하여인구집단에알맞은성별판별방정식의필요성을증명하고, 한국인머리뼈를이용한신원확인, 특히성별판별에유용한방정식을제시하는것에있다. 플릿을작성하여계측하였다 [12]. 3차원영상을사용한계측은뼈를직접계측하는것과는달리손으로만져뼈표지점을확인할수없기때문에정확한뼈표지점을찾기위해기준이될수있는평면 (plane) 을먼저설정하였다. 그평면은눈확위수평면 (Frankfort horizontal plane), 관상면 (frontal plane) 과시상면 (sagittal plane) 으로눈확위수평면을기준으로머리뼈의해부학적정렬을맞추고이면을기준으로나머지두개의면을만들어정확한뼈표지점을찾는것에사용하였다. 계측항목중정확한뼈표지점을찾기어려운 3개의항목들-위턱이틀길이 (maxilla-alveolar length; MaAl L), 아래턱가지최대높이 (maximum ramus height; Max Ra H), 아래턱길이 (mandibular length; Man L)-은방정식을이용하여계산하였다 [13]. 머리뼈 3차원영상을 3명의연구자가계측하였고계측한값은컴퓨터프로그램인 SPSS (version 20, IMB, USA) 를사용하여급내상관계수 (intraclass correlation coefficient; ICC), 수신자조작특성곡선 (receiver operating characteristic curve; ROC curve), 효과크기 (Cohen s d value), 대응표본 T검증 (paired t-test), 그리고판별분석 (discriminant function analysis) 등통계분석하였다. 판별분석은독립변수모두진입하는방법과단계선택법 (stepwise) 을 1) 44개항목모두, 2) 수신자조작특성곡선에서검증률이높게나온항목, 3) 수신자조작특성곡선과효과크기분석에서정확도와검증률이높은항목을사용하였다. 결과 재료및방법 한국인시신 CT 영상을제공하는 Digital Korean Human Model Database ( 중머리뼈에질환이나손상이없는남자 50개, 여자 52개의영상을 3차원재구성하였다. 사용한영상의연령분포는 21세에서 72 세이었고평균나이는 52세이었다. 계측항목은미국법과학회법의인류학과에서제시한항목들을사용하였으며 [12] 총 44개의계측항목중머리뼈 (cranium) 는 27개항목으로오른쪽과왼쪽을모두계측하는 3개항목이포함되어있고아래턱뼈 (mandible) 는양쪽을계측하는 7개항목을포함한 17개항목을계측하였다 (Table 1). 컴퓨터프로그램 Mimics (version 10.0, Materialise, Belgium) 를이용하여 CT 영상을 3차원재구성하였고같은프로그램내계측도구를이용하여표지점좌표기반의계측템 총 44개항목중 28개항목 ( 머리뼈 17개항목, 아래턱뼈 11개항목 ) 은성별에따라통계학적으로유의한차이가있었다 (P<0.05). 한쌍으로된 10개항목중 2개항목은오른쪽과왼쪽이통계학적으로유의한차이가있었다 (P<0.01) (Table 2). 머리뼈에서는 17개항목, 아래턱뼈에서는 6개의항목이급내상관계수 (ICC) 가 0.8 이상으로최소값은큰구멍너비 (foramen magnum breadth; FMa B) 로 0.008이었고최대값은 0.976으로큰구멍앞점-치조점길이 (basion-prosthion length; BaPr L) 이었다 (Table 3). 곡선아래영역 (AUC) 이 0.7 이상인항목은 16개, Cohen s d 값이 0.8 이상인항목은 8개이었고그중검증력이크다고할수있는 Cohen s d 값이 1 이상인것은 4개항목이었다 (Table 2 and Table 4). 높은정확도와검증력을나타내는 8개항목은 : 최대머리뼈길이 (maximum cranial length; Max CL), 최대머리뼈높이 (maximum cranial height; Max CH), 얼굴너비 (bizygomatic diameter; BiZ D), 머리뼈바닥길이 (cranial

3 머리뼈 3 차원영상을이용한성별판별 105 Table 1. Definition of cranial and mandibular measurements*. Part Abb Measurement Landmark Definition Cranium Max CL Maximum cranial length g-op Distance between glabella and opisthocranion in the midsagittal plane, measured in a straight line Max CB Maximum cranial breadth eu-eu Maximum width of skull perpendicular to midsagittal plane wherever it is located, with the exception of the inferior temporal lines and the area immediately surrounding them BiZ D Bizygomatic diameter zy-zy Direct distance between most lateral points on the zygomatic arch Max CH Maximum cranial height ba-b Direct distance from the lowest point on the anterior margin of foramen magnum to bregma Cb L Cranial base length ba-n Direct distance from nasion to basion BaPr L Basion-Prosthion length ba-pr Direct distance from basion to prosthion MaAl B Maxillo-alveolar breadth ecm-ecm Maximum breadth across the alveolar borders of the maxilla measured on the lateral surfaces at the location of the second maxillary molars MaAl L Maxillo-alveolar length pr-alv Direction distance from prosthion to alveolon BiA B Biauricular breadth au-au Least exterior breadth across the roots of the zygomatic processes UpF H Upper facial height n-pr Direct distance from nasion to prosthion Min FB Minimum frontal breadth ft-ft Direct distance between the two frontotemporale UpF B Upper facial breadth fmt-fmt Direct distance between the two external points on the frontomalar suture Na H Nasal height n-ns Direct distance from nasion to the midpoint of a line connecting the lowest points of the inferior margin of the nasal notches Na B Nasal breadth al-al Maximum breadth of the nasal aperture Or B Orbital breadth d-ec Laterally sloping distance from dacryon to ectoconchion Or H Orbital height Direct distance between the superior and inferior orbital margins BiOr B Biorbital breadth ec-ec Direct distance between right and left ectoconchion IntOr B Interorbital breadth d-d Direct distance between right and left dacryon FC Frontal chord n-b Direct distance from nasion to bregma taken in the midsagittal plane PC Parietal chord b-l Direct distance from bregma to lambda taken in the midsagittal plane OC Occipital chord l-o Direct distance from lambda to opisthion taken in the midsagittal plane FMa L Foramen magnum length ba-o Direct distance from basion to opisthion FMa B Foramen magnum breadth Distance between the lateral margins of foramen magnum at the points of greatest lateral curvature Ma L Mastoid length Vertical projection of the mastoid process below and perpendicular to the eye-ear plane Mandible Ch H Chin height id-gn Direct distance from infradentale to gnathion BH MeF Body height at mental foramen Direct distance from the alveolar process to the inferior border of the mandible perpendicular to the base at the level of the mental foramen BT MeF Body thickness at mental foramen Maximum breadth measured in the region of the mental foramen perpendicular to the long axis of the mandibular body BiG W Bigonial width go-go Direct distance between right and left gonion BiC B Bicondylar breadth cdl-cdl Direct distance between the most lateral points on the two condyles Min Ra B Minimum ramus breadth Least breadth of the mandibular ramus measured perpendicular to the height of the ramus Max Ra B Maximum ramus breadth Distance between the most anterior point on the mandibular ramus and a line connecting the most posterior point on the condyle and the angle of the jaw Max Ra H Maximum ramus height Direct distance from the highest point on the mandibular condyle to gonion Man L Mandibular length Distance of the anterior margin of the chin from a center point on the projected straight line placed along the posterior border of the two mandibular angles Man A Mandibular angle Angle formed by the inferior border of the corpus and the posterior border of the ramus *Most measurements are quoted from Standards for Data Collection from Human Skeletal Remains by Buikstra and Ubelaker (1994). This measurement has both sides.

4 106 김덕임, 이우영, 한승호 Table 2. Intraclass correlation coefficient (ICC) of cranial and mandibular measurements. Part Abb. Variable Cronbach s α ICC p-value Cranium Max CL Maximum cranial length <0.001 Max CB Maximum cranial breadth <0.001 BiZ D Bizygomatic diameter <0.001 Max CH Maximum cranial height <0.001 Cb L Cranial base length <0.001 BaPr L Basion-Prosthion length <0.001 MaAl B Maxillo-alveolar breadth <0.001 MaAl L Maxillo-alveolar length <0.001 BiA B Biauricular breadth <0.001 UpF H Upper facial height <0.001 Min FB Minimum frontal breadth <0.001 UpF B Upper facial breadth <0.001 Na H Nasal height <0.001 Na B Nasal breadth <0.001 Or B* Orbital breadth(left) <0.001 Orbital breadth(right) <0.001 Or H* Orbital height(left) <0.001 Orbital height(right) <0.001 BiOr B Biorbital breadth <0.001 IntOr B Interorbital breadth <0.001 FC Frontal chord <0.001 PC Parietal chord <0.001 OC Occipital chord <0.001 FMa L Foramen magnum length <0.001 FMa B Foramen magnum breadth Ma L* Mastoid length(left) <0.001 Mastoid length(right) <0.001 Mandible Ch H Chin height <0.001 BiG W Bigonial width <0.001 BiC B Bicondylar breadth <0.001 Min Ra B* Minimum ramus breadth (left) <0.001 Minimum ramus breadth (right) <0.001 Max Ra B* Maximum ramus breadth (left) <0.001 Maximum ramus breadth(right) <0.001 Max Ra H* Maximum ramus height(left) <0.001 Maximum ramus height (right) <0.001 Man L* Mandibular length(left) Mandibular length(right) <0.001 Man A* Mandibular angle(left) <0.001 Mandibular angle(right) <0.001 *This measurement has both sides. base length; Cb L), 위턱이틀너비 (maxilla-alveolar breadth; MaAl B), 오른쪽꼭지돌기길이 (right mastoid length; Ma L, right), 오른쪽턱뼈구멍에서몸통높이 (right body height at mental foramen; BH MeF, right), 턱뼈각너비 (bigonial width; BiG W) 이었다. 1. 모든항목적용항목의 AUC와 Cohen s d 값을고려하지않고 44개항목 모두를이용하여판별분석한결과단계선택법에의해머리뼈항목 6개, 아래턱뼈항목 4개로구성된방정식의교차유효값 (cross-validated accuracy) 이 86.7% 로가장높았다. 방정식은다음과같다. D =0.161 Ch H+ ( ) BH MeF (left) BH MeF (right) BiG W+ ( ) Max CB BiZ D + ( ) UpF B FC FMa B

5 머리뼈 3 차원영상을이용한성별판별 107 Table 3. Results from means, standard deviations, confidence intervals (CI) of difference, and Cohen s d between the sexes (unit: mm). Male (n=50) Female (n =52) Pooled (n =102) 95% CI of difference 95% CI for Cohen s d Variable p-value Cohen s d mean SD mean SD mean SD Lower Higher Lower Higher Max CL < Max CB BiZ D < Max CH < Cb L < BaPr L MaAl B < MaAl L BiA B UpF H Min FB UpF B Na H Na B Or B (left)* Or B (right)* Or H (left) Or H (right) BiOr B IntOr B FC < PC OC FMa L < FMa B Ma L (left)* Ma L (right)* < Ch H < BH MeF (left) BH MeF (right) < BT MeF (left) BT MeF (right)

6 108 김덕임, 이우영, 한승호 Table 3. Continued. Male (n=50) Female (n=52) Pooled (n =102) 95% CI of difference 95% CI for Cohen s d Variable p-value Cohen s d mean SD mean SD mean SD Lower Higher Lower Higher BiG W < BiC B Min Ra B (left) Min Ra B (right) Max Ra B (left) Max Ra B (right) Max Ra H (left) Max Ra H (right) < Man L (left) Man L (right) Man A(left) Man A(right) *This had both sides and showed statistically significant difference between left and right sides (P<0.01). Max Ra H (left) 이방정식의고유값은 1.818, 정준상관값은 0.803, Wilks 람다는 0.355, 유의확률은 보다작았다. 남자집단의정확도는 93.6%, 여자집단에서의정확도는 86.3% 로 89.8% 가올바로분류되었다. 2. AUC가 0.7 이상인항목만적용 AUC가중간이상인항목은머리뼈항목이 11개, 아래턱뼈항목이 5개이다. 16개항목을이용하여판별분석한결과머리뼈항목만으로단계선택법을사용한방정식 D4의교차유효값이 82.4% 로가장높았다 (Table 5). 16개항목을개별적으로판별분석하였을때 8개의항목의교차유효값이 70% 이상이었고가장높은항목은최대머리뼈높이 (Max CH) 와위턱이틀너비 (MaAl B) 로교차유효값이 74.5% 이었다. D4=0.291 BiZ D Cb L+ ( ) BiA B+ ( ) UpF B FC FMa B Ma L (right) AUC가 0.7 이상, Cohen s d 값이 0.8 이상인항목만적용 AUC가 0.7 이상이고 Cohen s d 값이 0.8 이상인항목은총 8개로이항목모두를사용한경우 (D7) 교차유효값은 76.5%, 단계선택법 (D8) 을사용한경우 77.6% 이었다 (Table 5). 8개의항목중최대머리뼈길이 (Max CL) 와오른쪽턱뼈구멍에서몸통높이 (BH MeF, right) 항목만각각 69.9% 로가장낮았고나머지 6개항목은모두 70% 이상이었다. D8=0.075 Max CH Ma L (right) BH MeF (right) BiG W AUC와 Cohen s d 값을이용하여분류한항목들로만판별분석하였을때교차유효값이가장큰방정식은 D4로성별판별의정확도는남자가 82.7%, 여자는 83.3% 이었다. 고찰 관찰자간오류 (interobserver error) 에서급내상관계수 (ICC) 가높은계측항목들은표지점의정의가명확하거나형태적특징파악이비교적용이한항목들이며, 낮은항목들은눈확 (orbit), 큰구멍 (foramen magnum), 그리고아래턱뼈아래모서리 (inferior margin of mandible) 와관련된항

7 머리뼈 3 차원영상을이용한성별판별 109 Table 4. Results of receiver operating characteristic (ROC) curve analysis. Variable AUC SD VRP Lower 95% CI of difference Higher Max CL Max CB BiZ D Max CH Cb L BaPr L MaAl B MaAl L BiA B UpF H Min FB UpF B Na H Na B Or B (left) Or B (right) Or H (left) Or H (right) BiOr B IntOr B FC PC OC FMa L FMa B Ma L (left) Ma L (right) Ch H BH MeF (left) BH MeF (right) BT MeF (left) BT MeF (right) BiG W BiC B Min Ra B (left) Min Ra B (right) Max Ra B (left) Max Ra B (right) Max Ra H (left) Max Ra H (right) Man L (left) Man L (right) Man A(left) Man A(right) 목들이었다 (Table 3). 수신자조작특성곡선 (ROC curve) 은항목의정확도를사정하는것으로곡선아래영역 (AUC) 값을이용하여정보의정확도를나타낸다 [14]. Cohen s d 값은가설검증에서통계적검증력에영향을주는요인으로그값이 0.8 이면검증력이크다고설명할수있다 [15]. ICC 가 0.8 이상이고 AUC가 0.7 이상, Cohen s d 값이 0.8 이상인항목은총 44개항목중 8개항목으로이것은서로다른연구자가계측해도오차의범위가무시할수있을정도로작으며, 항목이남녀성별판별에높은정확도를가지고있다고할수있다 [2,14,15].

8 110 김덕임, 이우영, 한승호 Table 5. Discriminant function analysis using 3-dimensional images of skull. Functions Coefficient Standardized Unstandardized Eigenvalue Canonical correlation Wilk s Lambda Sectioning point Original Accuracy(%) Males Females Pooled Cross-validated D. All variables_stepwise* Ch H BH MeF (left) BH MeF (right) BiG W Max CB BiZ D UpF B FC FMa B Max Ra H (left) (Constant) D1. All variables_direct Max Ra H (left) Max Ra H (right) BiG W BiC B BH MeF (right) Max CL BiZ D Max CH Cb L MaAl B BiA B UpF B FC FMa L FMa B Ma L (right) (Constant) D2. All variables_stepwise BiG W BH MeF (right)

9 머리뼈 3 차원영상을이용한성별판별 111 Table 5. Continued. Functions Coefficient Standardized Unstandardized Eigenvalue Canonical correlation Wilk s Lambda Sectioning point Original Accuracy(%) Males Females Pooled Cross-validated FC FMa B Ma L (right) (Constant) D3. Cranium_direct Max CL BiZ D Max CH Cb L MaAl B BiA B UpF B FC FMa L FMa B Ma L (right) (Constant) D4. Cranium_stepwise BiZ D Cb L BiA B UpF B FC FMa B Ma L (right) (Constant) D5. Mandible_direct BH MeF (right) BiG W BiC B Max Ra H (left) Max Ra H (right) (Constant)

10 112 김덕임, 이우영, 한승호 Table 5. Continued. Functions Coefficient Standardized Unstandardized Eigenvalue Canonical correlation Wilk s Lambda Sectioning point Original Accuracy(%) Males Females Pooled Cross-validated D6. Mandible_stepwise BH MeF (right) BiG W BiC B Max Ra H (left) (Constant) D7. Cranium + Mandible_direct Max CL Max CH BiZ D Cb L MaAl B Ma L (right) BH MeF (right) BiG W (Constant) D8. Cranium + Mandible_stepwise Max CH Ma L (right) BH MeF (right) BiG W (Constant) *This equation was not effected by AUC and Cohen s d value and took advantage of 44 variables. Two equations, D7 and D8, were built up variables that were sorted by AUC (> 0.7) and Cohen s d value (> 0.8).

11 머리뼈 3 차원영상을이용한성별판별 113 Table 6. Comparison of mean values between this study and other population groups (unit: mm) This study (2015) Franklin et al. (2013) Ogawa et al. (2013) Guyomarch & Bruzek (2011) Saini et al. (2011) Dayal et al. (2008) Steyn & İşcan (1998) Population Korean W. Australian Japanese French Thai Indian S. African Black S. African White Materials 3D image 3D image Bone Bone Bone Bone Bone Bone Sex (N) M (50) F (52) M (190) F (190) M (73) F (40) M (24) F (21) M (46) F (44) M (82) F (30) M (60) F (60) M (43) F (46) Max CL ±6.72 Max CB ±6.90 BiZ D ±5.39 Max CH ±4.37 Cb L ±4.26 BaPr L ±4.72 MaAl B ±3.01 BiA B ±5.50 UpF H ±3.28 Min FB ±4.33 UpF B ±4.45 Na H ±3.11 Na B ±1.67 Or B ±1.71 Or H ±1.81 BiOr B ±4.40 IntOr B ±2.35 FC ±7.54 PC ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±5.50* 62.2 ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±4.74* ± ±7.04* ± ± ± ± ± ±6.8* ±7.8* ± ± ± ± ± ± ± ± ± ± ±7.3* ± ± ± ± ± ± ± ±5.5* ±7.6* ± ± ± ± ± ± ±4.7* 92.1 ± ±1.9* ± ±5.7* ±8.2* ± ± ± ± ±4.6* 66.8 ±4.3* 91.7 ± ±2.2* ±5.3* ±6.6* ± ± ± ± ± ± ± ± ± ± ±3.40* ± ± ±1.70* ±1.17* ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ± ±4.25* ± ± ± ± ± ± ± ± ±5.39* 71.3 ±3.91* 53.7 ± ± ± ± ± ± ± ±5.13* 49.8 ± ±2.06

12 114 김덕임, 이우영, 한승호 Table 6. Continued. Steyn & İşcan (1998) Dayal et al. (2008) Saini et al. (2011) Guyomarch & Bruzek (2011) Ogawa et al. (2013) Franklin et al. (2013) This study (2015) Population Korean W. Australian Japanese French Thai Indian S. African Black S. African White Materials 3D image 3D image Bone Bone Bone Bone Bone Bone Sex (N) M (50) F (52) M (190) F (190) M (73) F (40) M (24) F (21) M (46) F (44) M (82) F (30) M (60) F (60) M (43) F (46) 97.1 ± ± ± ± ± ± ± ± ± ±3.54 OC ± ± ±3.39* 39.3 ± ± ± ± ± ± ± ± ± ± ± ± ±5.13* ± ±5.66* ±4.88 FMa L ±1.90 Ma L ±3.80 BiG W ±5.65 BiC B ±12.6 Min Ra B ±2.90 * That was not showed statistically significant difference between population groups (P>0.05), the rest was showed statistically significant difference among population groups (P<0.05). 머리뼈와골반뼈는성별판별에있어서높은정확도를가지고있으며머리뼈만으로도남녀성별판별의정확도는 80~92%, 골반뼈만사용한경우정확도는 90~95% 이며두개의뼈를모두사용할경우 90~100% 의정확도를보인다 [3,16,17]. 또한인구집단간그특징을잘반영하고있는뼈들중가장넓게연구되고있는것이머리뼈이다 [18]. 한국인뼈를사용한성별판별연구들의결과를살펴보면목뿔뼈 88.2% [1], 발꿈치뼈 65.4~89.4% [2], 넙다리뼈의안쪽 가쪽관절융기 72.3~90.1% [5], 골반뼈는 60% 이상 [16], 목말뼈 67.1~87.1% [19], 고리뼈 93.9% [20], 정강뼈 84.8% [21], 그리고아래턱뼈를이용한판별방정식의정확도는 50.4~88.8% 이었다 [22,23]. 이와같이성별을판별할수있는방정식을제공하기위해여러뼈들을이용하여계측값을판별분석한연구결과는있으나높은정확도를가지고있다는머리뼈의경우한국인을대상으로한판별방정식은없었다. 한국인머리뼈 3차원영상을이용한이연구에서성별판별분석의교차유효값은최소 76.5% 에서최대 86.7% 이었다 (Table 5). 한국인 3차원머리뼈영상계측값과다른인구집단의머리뼈계측값을비교한결과유럽인 [24,25], 동양인 [17,25,26], 그리고아프리카흑인과백인 [27,28] 과는일치하는항목의대부분이통계학적으로유의한차이가있었다 (P<0.05) (Table 6). 3차원머리뼈영상을이용한 Franklin 등 [24] 의연구와일치하는항목 10개모두여자는통계학적으로유의한차이가있었으나 (P<0.05), 남자는큰구멍앞점-치조점길이 (BaPr L) 항목만차이가없었다. 한국인여자머리뼈는오스트레일리아인의계측항목과일치하는 10개모두통계학적으로유의한차이가있었으며, 남자머리뼈는인도인 (9개항목 ) 과아프리카인 (12개항목 ) 과일치하는항목모두통계학적으로유의한차이가있었다 (Table 6). 통계학적으로유의한차이가없는항목들을살펴보면남자는인구집단간공통된항목이없으나여자는동양인과의비교에서최대머리뼈길이 (Max CL), 큰구멍앞점-치조점길이 (BaPr L), 그리고코너비 (nasal breadth; Na B) 는차이가통계학적으로유의하지않았다. 한국인여자의경우머리뼈길이가일본인, 태국인과비슷하고뼈콧구멍 (aperture piriformis) 의너비는태국인, 인도인과비슷하였다. 머리뼈를이용한판별분석결과를비교해보면정확도가가장높은판별방정식은이연구를비롯해서다른인구집단의연구에서도모두머리뼈 (cranium) 항목으로구성되어있으며방정식을구성하는항목이최소 1개에서최대 7개로구성되어있었다 (Table 7). 이연구의결과와비교한 6개논문의 7개방정식을구성하고있는항목들중가장많이사용된것은얼굴너비 (BiZ D), 최대머리뼈길이 (Max CL),

13 머리뼈 3 차원영상을이용한성별판별 115 Table 7. Comparison of discriminant function equation using cranium between this study and other population groups. Method Variable Unstandardized coefficient Wilk s Lambda Sectioning point Accuracy (%) Original group Cross-validated group This study (2015) Stepwise BiZ D Cb L BiA B UpF B FC FMa B Ma L (right) Constant Franklin et al. (2013) Stepwise Max CL BiZ D Ma L (right) Constant Ogawa et al. (2013) Stepwise Max CL Cb L Max. Frontal B* UpF B Constant Guyomarch & Bruzek (2011): French Giles & Elliot (1963) Max CL BaPr L BiZ D UpF H Guyomarch & Bruzek (2011): Thai Henke (1973) Max CH Cb L Saini et al. (2011) Stepwise BiZ D Constant Dayal et al. (2008) Stepwise Max CL Max CH BiZ D Na H Constant Steyn & İşcan (1998) Stepwise Max CL BiZ D Cb L Max CH Na H Na B Constant *That was not accorded with variables of this study.

14 116 김덕임, 이우영, 한승호 Table 8. Comparison of accuracy using equations from other studies. Study Population group Equation Accuracy (%) Male Female Pooled This study (2015) Korean D Franklin et al. (2013) W. Australian Function Ogawa et al. (2013)* Japanese Function Saini et al. (2011) India Function *Function 1 in Ogawa study was the best equation but one variable was not matched of this study. Function 4 was the second best equation to determine sex in Japanese and all variables in Function 4 were matched this study. 머리뼈바닥길이 (Cb L), 그리고최대머리뼈높이 (Max CH) 순이다. 이연구와비교한 6개의논문중 3차원영상을이용한 Franklin 등의논문 [24] 과일본인을대상으로한논문 [17], 그리고하나의항목만으로높은정확도를가진방정식 [26] 을이용하여이연구에서계측한값을대입하여성별판별의정확도를확인하였다 (Table 8). 유럽인을대상으로만들어진방정식에이연구의계측값을이용한경우약 76% 의정확도를보였으며남자보다여자가더정확하게판별되었고같은방법으로일본인을대상으로만들어진방정식으로는남자가더정확하게판별되었다. 인도인을대상으로한하나의항목으로구성된방정식은판별의정확도가 85.5% 이었고방정식에계측값을넣어계산한결과한국인남자는모두남자로구분되었으나여자의경우 7.7% 의정확도를나타내어여자머리뼈대부분을남자의머리뼈로잘못구분하였다. 한국인얼굴너비 (BiZ D) 의평균값은인도인의평균값과비교했을때통계학적으로유의한차이를보였으며특히한국인여자의얼굴너비평균값이인도인남자의평균값보다커여자의대부분이남자로판별된것으로생각된다. 이와같은차이는 Franklin 등 [24] 은이연구와같은 3차원영상을이용하였고 Ogawa 등 [17] 은우리나라와지리학적으로가까운나라이기때문에그들의방정식을이용하였을경우에도남자와여자의판별정도가비슷하였으나 Saini 등 [26] 은유럽인에가까운인구집단인북인도인의뼈를이용하였기때문일것이라생각한다. 한국인머리뼈 3차원영상을이용한남녀성별판별은교차유효값이 76.5~86.2% 이었다. 다른인구집단의계측값과비교하였을때거의모든항목이통계학적으로유의한차이가있었으며유럽인, 동양인, 아프리카인을대상으로한판별방정식에한국인머리뼈계측값을이용하여남녀판별을한결과남자는유럽인의머리뼈보다작았고여자는일본인, 인도인보다머리뼈가크다는것을추측할수있었다. 이러한연구의결과로특정인구집단에알맞은방정식이필요함을제안할수있다. 이연구에서도출된방정식은추후현장에서발견된한국인머리뼈 3차원영상을이용한성별판별에도움을줄것이라고생각한다. 추후인구집단간특 징을나타내는항목과그항목의한국인기준값을도출해낸다면좀더쉽고정확한인구집단별성별을확인할수있을것이다. 참고문헌 1. Kim DI, Lee UY, Park DK, Kim YS, Han KH, Kim KH, et al. Morphometrics of the hyoid bone for human sex determination from digital photographs. J Forensic Sci. 2006; 51: Kim DI, Kim YS, Lee UY, Han SH. Sex determination from calcaneus in Korean using discriminant analysis. Forensic Sci Int. 2013; 228:177e.1-177e Thompson T, Black S. Forensic Human Identification, An introduction, Boca Raton, CRC Press, pp , Byers SN. Introduction to Forensic Anthropology, 4 th ed., Boston, Pearson, pp , Kim DI, Kwak DS, Han SH. Sex determination using discriminant analysis of the medial and lateral condyles of the femur in Koreans. Forensic Sci Int. 2013; 233: Han SH, Hwang YI, Lee KH, Koh KS, Choi BY, Lee KS, et al. Craniometric study in modern Korean adults. Korean J Phys Anthropol. 1995; 8: Kim HJ, Kim KD, Choi JH, Hu KS, Oh HJ, Kang MK, et al. Differences in the metric dimensions of craniofacial structures with aging in Korean males and females. Korean J Phys Anthropol. 1998; 11: White TD, Folkens PA. The human bone manual, Burlington, Elsevier, pp 385-7, Kim DI, Lee SS, Kim YS. Statistical analysis of bone elements excavated from the forensic context. Korean J Phys Anthropol. 2010; 23: Kim DI, Lee UY, Park SO, Kwak DS, Han SH. Identification using frontal sinus by three-dimensional reconstruction from computed tomography. J Forensic Sci. 2013; 58: Park SH, Yu HS, Kim KD, Lee KJ, Baik HS. A proposal for a new analysis of craniofacial morphology by 3-dimensional computed tomography. Am J Orthod Dentofacial Orthop.

15 머리뼈 3 차원영상을이용한성별판별 ; 129:600.e e Buikstra JE, Ubelaker DH Standards for Data Collection from Human Skeletal Remains. Arkansas Archaeological Survey Research Series No. 44. Fayetteville. 13. Lee UY, Park DK, Kim YS, Lee SS, Kwak DS, Kim DH, et al. Construction of 3 dimensional head models for forensic anthropology. Oral presentation 26. World Forensic Festival Seoul, p133, Swets JA. Measuring the accuracy of diagnostic systems. Science. 1988; 240: Wikimedia Foundation, InC.. Wikipedia. Effect size [Internet]. Available from: size 16. Choi BY, Chung IH. Sex discrimination with the metric measurements of the Korean dried pelvic bones by discriminant function analysis. Korean J Phys Anthropol. 1999; 12: Ogawa Y, Imaizumi K, Miyasaka S, Yoshino M. Discriminant functions for sex estimation of modern Japanese skulls. J Forensic Leg Med. 2013; 20: Perera P, Pathmeswaran A. A Pilot study on assessment of racial affinity of Sri Lankan population using discriminant function statistics and a few established morphological racial traits. Leg Med. 2009; 11:S Lee UY, Han SH, Park DK, Kim YS, Kim DI, Chung IH, et al. Sex determination from the talus of Koreans by discriminant function analysis. J Forensic Sci. 2012; 57: Park DK, Ra JJ, Park KH, Ko JS, Kim DI, Kim YS, et al. Determination of sex in Koreans using atlas. Korean J Phys Anthropol. 2009; 22: Lee JH, Han HS, Chung IH. Sex determination from the tibia in Korean population. Korean J Phys Anthropol. 2010; 23: Lin C, Jiao B, Liu S, Guan F, Chung NE, Han SH, et al. Sex determination from the mandibular ramus flexure of Koreans by discrimination function analysis using three-dimensional mandible models. Forensic Sci Int. 2014; 236: 191.e1-191.e Hu KS, Koh KS, Jung HS, Kang MK, Choi BY, Kim HJ. Physical anthropological characteristics and sex determinative analysis by the metric traits of Korean mandibles. Korean J Phys Anthrop. 2000; 13: Franklin D, Cardini A, Flavel A, Kuliukas A. Estimation of sex from cranial measurements in a Western Australian population. Forensic Sci Int. 2013; 229:158.e1-158.e Guyomarc h P, Bruzek J. Accuracy and reliability in sex determination from skulls: A comparison of Fordisc 3.0 and the discriminant function analysis. Forensic Sci Int. 2011; 208:180.e1-180.e Saini V, Srivastava R, Rai RK, Shamal SN, Singh TB, Tripathi SK. An osteometric study of northern Indian populations for sex dimorphism in craniofacial region. J Forensic Sci. 2011; 56: Dayal MR, Spocter MA, Bidmos MA. An assessment of sex using the skull of black South Africans by discriminant function analysis. HOMO. 2008; 59: Steyn M, İşcan Y. Sexual dimorphism in the crania and mandibles of South African whites. Forensic Sci Int. 1998; 98:9-16.

16 118 김덕임, 이우영, 한승호 Sex Determination Using Three-dimensional Image of Skull in Korean: Metric Study by Discriminant Function Analysis Deog-Im Kim 1, U-Young Lee 2, Seung-Ho Han 3 1 Department of Nursing, College of Nursing, Keimyung University 2 Department of Anatomy Catholic Institute for Applied Anatomy, College of Medicine, The Catholic University of Korea 3 Department of Anatomy, College of Medicine, Chung-Ang University Abstract : Sex determination is considered the first and most important process in the identification of unknown skeletal remains. The skull is one of bones with high accuracy to discriminate sexes, but there is no dimorphic studies of Korean skull by discriminant function analysis. The aim of this study was to build and provide an easy and accurate discriminant equation to sex determination by using three-dimensional skull images of Korean. Computed tomography images from 102 cadavers were reconstructed three-dimensional images by computer program. We measured 44 variables using the template in computer program and variables were sorted out items with high accuracy and reliability by intraclass correlation coefficient (ICC), Cohen s d value, and receiver operating characteristic (ROC) curve. The equation with the highest accuracy had 82.7% in males, 82.2% in females, it was constituted bizygomatic diameter, cranial base length, biauricular breadth, upper facial breadth, frontal chord, foramen magnum breadth, and right mastoid length. The cross-validated accuracy had 76.5%~86.7% using 3D skull images in Koreans. The value of variables that matches the other population group study, most of variables had a statistically significant difference among population groups. Male skulls in Koreans were smaller than those in European and female skulls were bigger than in Asian. This result should be helpful to determine sex in Koreans. Keywords: Three-dimensional images, Skull, Discriminant function analysis, Metric study, Korean Correspondence to : Seung-Ho Han (Department of Anatomy, College of Medicine, Chung-Ang University) monami@cau.ac.kr