Computer-aided system for morphometric mandibular index computation (Using dental panoramic radiographs)

Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e624 Jou nal sec ion: Ge odon ology Publica ion Types: Re iew Compu e -aided sys em o mo phome ic mandibula index compu a ion (Using den al pano amic adiog aphs) Jose López-López 1, Jose-Manuel Ál a ez-López 2, En ique Jané-Salas 1, Albe Es ugo-De esa 3, Raul Ayu- so-Mon e o 4, Eugenio Velasco-O ega 5, Juan-Jose Segu a-Egea 6 1 PhD. MD, DDS. Specialis in S oma ology. School o Den is y. Depa men o Odon o-s oma ology, Uni e si y o Ba celona 2 PhD, DDS. Compu e Science esea ch scien is , CBLL, New Yo k Uni e si y, and Compu e Vision Cen e , Ba celona 1 PhD. MD, DDS. Specialis in S oma ology. School o Den is y. Depa men o Odon o-s oma ology, Uni e si y o Ba celona 3 MD, DDS. Specialis in S oma ology. School o Den is y. Depa men o Odon o-s oma ology, Uni e si y o Ba celona 4 PhD, DDS. School o Den is y. Depa men o Odon o-s oma ology, Uni e si y o Ba celona 5 PhD, MD, DDS. Specialis in S oma ology. School o Den is y. Depa men o S oma ology, Uni e si y o Se illa 6 PhD, MD, DDS. Specialis in S oma ology. School o Den is y. Depa men o Endodon ics, Uni e si y o Se illa Co espondence: Depa amen o Odon o-S oma ology Facul y o Odon ology-Uni e si y o Ba celona Uni e si a y Campus o Bell i ge C/ Feixa LLa ga S/N 08907 L’Hospi ale de Llob ega (Ba celona, Spain) [email protected] Recei ed: 03/04/2011 Accep ed: 16/09/2011 Abs ac Objec i e: We p opose and alida e a compu e —aided sys em o measu e h ee di e en mandibula indexes: co ical wid h, pano amic mandibula index and, mandibula al eola bone eso p ion index. S udy Design: Repea abili y and ep oducibili y o he measu emen s a e analyzed and compa ed o he manual es ima ion o he same indexes. Resul s: The p oposed compu e ized sys em exhibi s supe io epea abili y and ep oducibili y a es compa ed o s anda d manual me hods. Mo eo e , he ime equi ed o pe o m he measu emen s using he p oposed me hod is negligible compa ed o pe o m he measu emen s manually. Conclusions: We ha e p oposed a e y use iendly compu e ized me hod o measu e h ee di e en mo phome - ic mandibula indexes. F om he esul s we can conclude ha he sys em p o ides a p ac ical manne o pe o m hese measu emen s. I does no equi e an expe examine and does no ake mo e han 16 seconds pe analysis. Thus, i may be sui able o diagnose os eopo osis using den al pano amic adiog aphs. Key wo ds: Os eopo osis, pano amic mandibula index, co ical wid h, mandibula al eola bone eso p ion index. López-López J, Ál a ez-López JM, Jané-Salas E, Es ugo-De esa A, Ayuso-Mon e o R, Velasco-O ega E, Segu a-Egea JJ. Compu e -aided sys em o mo phome ic mandibula index compu a ion (Using den al pano amic adiog aphs). Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. h p://www.medicinao al.com/medo al ee01/ 17i4/medo al 17i4p624.pd A icle Numbe : 17637 h p://www.medicinao al.com/ © Medicina O al S. L. C.I.F. B 96689336 - pISSN 1698-4447 - eISSN: 1698-6946 eMail: [email p o ec ed] Indexed in: Science Ci a ion Index Expanded Jou nal Ci a ion Repo s Index Medicus, MEDLINE, PubMed Scopus, Embase and Emca e Indice Médico Español doi:10.4317/medo al.17637 h p://dx.doi.o g/doi:10.4317/medo al.17637 Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e625 In oduc ion Os eopo osis is a sys emic bone disease cha ac e ized by educ ion in bone mass and mic o-a chi ec u al de- e io a ion o bone issue. Os eopo osis is he main dis- ease o middle age women, especially menopausal ones. I a ec s o mo e han 75 million people in Eu ope, Ja- pan and he Uni ed S a es. Spain has one o he highes incidences in Eu ope since he e a e mo e han 2.5 mil- lions o a ec ed woman. Fu he , 40% o hese women a e no diagnosed ye (1). Al hough mass bone and cal- cium me abolism al e a ions a e e iden in he p emen- opausal pe iod, he menopause s esses he beginning o bone loss. I is p olonged un il li e’s end and i is he main cause o bone ac u es in olde age women (2,3). The Spanish Associa ion Agains Os eopo osis (AE- COS) es ima es an annual hip ac u e a e o 60.000 in Spain. Fu he , he annual a e in he Eu opean Union is expec ed o inc ease om 414.000 o 972.000 a ec ed people wi hin he nex 50 yea s. P e en i e measu es and an ea ly diagnose may help dec easing hese a es signi ican ly (4). The ea ly onse o os eopo osis can be de ec ed by dec eased densi y o no mal bone. The mos widely used echnique o bone densi y es ing is dual ene gy X- ay abso p iome y (DXA) (5) consis ing o wo x- ay beams wi h di e ing ene gy le els aimed a he pa ien ’s bones. Bone densi y da a is in e p e - ed in compa ison wi h wha is conside ed no mal o a heal hy young pe son. In his way, he bone densi y es s calcula e he pe son’s T-sco e. Then, i is compa ed agains op imal bone densi y and exp essed as numbe o s anda d de ia ions below he a e age. Finally, ac- co ding o Wo ld Heal h O ganiza ion c i e ia (6), a T- sco e g ea e han 1 is conside ed no mal. A T-sco e o less han -2.5 is indica i e o os eopo osis and a T-sco e anging om -1 o -2.5 is conside ed as a condi ion o os eopenia ( he p ecu so o os eopo osis). Recen s udies indica e he possibili y o assessing he bone mine al densi y (DMO) analyzing he mandibula bone mine al densi y (7). In addi ion, some au ho s e- la e os eopo osis o he mandibula esidual idge eso p- ion, pe iodon al bone loss and oo h loss (8,9). These s udies assess and co ela e he os eopo osis diagnose wi h he adiog aphic esul s in pano amic adiog aphs (o opan omog aphs, OPG) and pe iapical adiog aphs. Fu he , hey highligh he ele ance o he den is in he ea ly onse o his disease. The main den al pano amic adiog aphy measu es o he os eopo osis diagnose a e: co ical wid h (CW), also known as mandibula co i- cal hickness (MTC) (10), pano amic mandibula index (PMI) (11) and, mandibula al eola bone eso p ion in- dex (MM a io) (12); see (Fig. 1). Howe e , he alidi y o hese indexes and ela ed ones such as mandibula angle (13) and he isual mo phology o he co ical (14) is unde e iew. Mo eo e , he p ecise measu emen o hese indexes, in e ms o epea abili y and ep oduc- ibili y, on den al pano amic adiog aphs is no p ac ical Fig. 1. Main measu es a e de ined as: a = Co ical wid h (CW); a/b= pano amic mandibula index (PMI); d/c= mandibula al eola bone eso p ion index (MM a io). Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e626 and is complica ed in clinical den al p ac ice (14-16). These indexes ha e been widely used in he li e a u e o e alua e he DMO and, hus, o diagnose os eopo o- sis. Fu he , conside ing he o hopan omog aphy is a common p ac ice in den al clinics mos au ho s p o- pose using den al adiog aphy measu emen s o he ea ly diagnose o os eopo osis. Tha is, hey use hese measu emen s o ad ise pa ien s o de ailed s udies o hei bone pa hology. Fo ins ance, Ho ne K & De lin H (7) in 1998 conclude ha he DMO assessed by DXA is signi ican ly co ela ed wi h DMO measu emen s us- ing MTC and PMI in pano amic adiog aphs. Fu he , in 2002, he same au ho s use measu emen s o co ical hickness a he den al mandibula o amen (mandibula index, MI), an egonion (an egonial index, AI) and go- nion (gonial index, GI) egions o conclude ha only MI con ibu ed signi ican ly o a diagnosis o low skele al DMO (17). Whi e SC e al. (18) in 2005, use MTC and isual co ex analysis desc ibed in Kleme i e al. (14) o conclude ha bo h indexes a e use ul o inden i y- ing subjec s ha ing low bone mass. A i in AZ e al. (19) in 2006, p esen a compu e -aided me hod o es ima e MTC in digi al adiog aphs and o assess he DMO o he pa ien s. Taguchi e al. (20) in 2006, use MTC and he isual co ex analysis combined wi h a isk ac o s su ey ( he os eopo osis sel -assessmen ool) gi en o 150 pa ien s o conclude ha den is s ha e su icien clinical and adiog aphic in o ma ion o ad ise women (younge han 65 yea s old) o unde go a bone densi- ome y analysis. In ano he s udy, in 2007, Taguchi e al. (21) use MTC and he isual co ex analysis o conclude ha pos menopausal women wi h mandibula dis o ions usually ha e low DMO o e en os eopo osis. Mo e ecen ly, Vlasiadis e al. (22) in 2008, use men al pano amic adiog aphy indexes as a simple me hod in he diagnose o os eopo osis. They conclude ha he e is a ela ionship be ween spinal BMD, mandibula co - ical wid h and Kleme i isual co ex analysis. The e o e, gi en he di icul y associa ed o he meas- u emen o men al pano amic adiog aphy indexes we p opose a compu e -aided sys em o compu e hese measu es exhibi ing, a leas , he same e iciency han manual (cu en ) es ima ion. Hence, he pu pose o his s udy is alida ing he bene i s o es ima ing he meas- u es using he p oposed sys em in on o he manual me hod. Tha is, we aim o p o e he use ulness o us- ing a compu e -aided sys em o es ima e mo phome ic measu es in o hopan omog aphy. Subjec and Me hods Ten pano amic adiog aphs om 10 di e en subjec s a e andomly selec ed om a se o 200 adiog aphs aken a he Uni e si y Den al Clinic, Uni e si y o Ba celona. The subse o adiog aphs is analyzed by 10 di e en examine s. They a e den is s wi h mo e han 3 yea s o clinical expe ience. All o hem ook a 1 hou heo y-and-p ac ice cou se o lea n he manual and compu e -aided measu emen p ocedu es. Measu es o e alua e: Th ee di e en measu es a e e alua ed: co ical wid h (CW), pano amic mandibula index (PMI) and, man- dibula al eola bone eso p ion index (MM a io) (Fig. 1). Manual measu emen o hese indexes is ca ied ou d awing lines pa allel o he long axis o he mandible and angen ial o he in e io bo de o he mandible. Fu he , a line pe pendicula o his angen in e sec - ing he in e io ma gin o he men al o amen is d awn. Measu emen s a e aken along his pe pendicula line as shown in (Fig. 1). I he men al o amen is isible on bo h sides o he mandible he measu emen s a e a e - aged. Tha is, he mean alue o he indexes is consid- e ed. Al hough his is a simple p ocess o es ima e he measu emen s i is no p ac ical p o iding high in e - obse e a ia ions. Tha is, measu emen s p o ided by di e en den is s a e di e en . Compu e -aided measu emen sys em: In his sec ion, he p oposed algo i hm o es ima ing mo phome ic mandibula indexes is in oduced. The algo i hm is based on he knowledge o se e al poin s manually anno a ed in he image (Fig. 2A). These poin s de ine he uppe and lowe bo de o he mandible, he in e io ma gin and he cen e o he men al o amen, and he co ical wid h. Thus, he edious p oblem o lo- ca ing he mandibula axis and measu ing dis ances in a p in ed o hopan omog aphy is educed o selec nine poin s in a digi al image. Based on hese poin s he algo i hm d aws suppo lines o es ima e he mo phome ic mandibula indexes. In his way, he uppe and lowe bo de s a e modeled using second o de cu es (pa abolas). Using hese pa abolic models has wo main ad an ages: i s simplici y since only h ee adjus able coe icien s a e equi ed, and i s physical plausibili y since he bo de o he mandible can be easonably app oxima ely by a pa abolic unc- ion in he image plane (Fig. 2B). The coe icien s o hese models a e ixed based on inpu poin s (Pmu and Pml espec i ely) in a leas squa e sense. Tha is, he coe icien s o each cu e minimize he sum o squa ed esiduals. Residuals a e de ined as he sho es Euclid- ian dis ance be ween each inpu poin and he esul ing cu e. Gi en he lowe bo de o he mandible and he poin de ining he in e io ma gin o he men al o amen P l, he suppo line ℓÞ passing pe pendicula o he bo de and h ough P l is d awn. This line de ines he sho - es dis ance be ween he in e io ma gin o he o amen and he lowe bo de o he mandible. Then, he slope o he mandibula axis is uniquely de ined by he line ℓß angen o he lowe mandibula bo de a xo, he in- e sec ion poin be ween ℓÞ and he lowe mandibula Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e627 bo de (Fig. 2C). De i a i es o he cu e a e aken and e alua ed a xo o de i e he slope o ℓß. Then, he slope o ℓÞ is ob ained since hese wo lines a e pe pendicu- la . Robus ness agains in e io o amen ma gin de ini- ion is achie ed by selec ing mul iple poin s de ining he in e io ma gin o he men al o amen. In his case, he algo i hm selec s he poin co esponding o he mini- mum dis ance. A his poin , co ical wid h (C) is es ima ed along ℓÞ using Pcw, he co ical wid h poin . This poin loca es he in e sec ion be ween he in e io mandibula co ex and ℓÞ. Fu he mo e, he pano amic mandibula index (PMI) is compu ed as he a io o he hickness o he in e io mandibula co ex in he men al egion o e he dis ance be ween he lowe bo de o he mandible and he in e io ma gin o he men al o amen. This index is also compu ed along ℓÞ using Pcw. The emaining is es ima ing he M/M a io. This is done ex ending ℓÞ o in e sec he uppe bo de o he mandi- ble as shown in (Fig. 2D). Measu es a e aken along his line be ween mandible bo de s and he lowe bo de and P c, he cen e o he men al o amen. I mo e han one poin is selec ed o de ine he cen e o he o amen hei cen oid is used o compu e he measu e. All he measu es used o compu e he indexes a e gi en in pixels. Ne e heless, he ou pu o he algo i hm is gi en in millime e s using wo con e sion a ios: dis o - ion a io and pixel o millime e a io. The o me e e s o he dis o ion in oduced in he imaging p ocess. The la e e e s o he dis ance o he mandible co e ed by each pixel in he image. These a ios a e de ice depend- en . The dis o ion a io mus be in oduced manually whe eas he pixel o millime e a io can be es ima ed by compa ing manually measu ed dis ances in a p in ed o hopan omog aphy agains he numbe o pixels in he co esponding image. Finally, he algo i hm o compu ing he mo phome ic mandibula indexes is summa ized as ollows: 1) Pe - o m algo i hm calib a ion i equi ed. 2) Load and dis- play he egion o in e es o an o hopan omog aphy. 3) Selec a leas h ee poin s de ining he uppe bo de o he mandible (Pmu). 4) Selec a leas h ee poin s de in- ing he lowe bo de o he mandible (Pml). 5) Selec a leas one poin de ining he in e io ma gin o he men- al o amen (P l). 6) Selec a leas one poin de ining he cen e o he men al o amen (P c). 7) Selec he poin (Pcw) along he line ℓÞ o de ine he mandibula in e io co ical wid h. This line is angen o he lowe mandi- Fig. 2A. The algo i hm is based on nine poin s manually labelled in each image. Pmu: poin s de ining he uppe mandibula bo de . Pml: poin s de ining he lowe mandibula bo de . P c: he cen e o he o amen. P l: in e io ma gin o he o amen. Pcw: poin s de ining he co ical wid h. 2B. Second o de cu es (pa abolas) a e used o model he uppe and lowe bo de s o he mandible. These cu es can easonably app oxima e he eal shape o he mandible based on a leas h ee poin s. 2C. Suppo lines a e d awn o de ine he slope and posi ion o he mandibula axis. 2D. Di e en suppo lines a e d awn au oma ically based on use inpu . ℓÞ is ex ended o in e sec he uppe bo de o he mandible. All he indices a e measu ed along his ex ension line. Co ical wid h is de ined by a, Pano amic mandibula index is a/b and he M/M a io is d/c. Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e628 ble bo de and passes h ough he in e io o amen ma - gin. 8) Sa e he esul s i equi ed. Me hodology: Th ee di e en measu emen s se s a e ca ied ou by each examine in each o hopan omog aphy wi h en minu es b eak be ween measu emen s. OPG a e an- domly selec ed a each measu emen ound. This p oc- ess is epea ed using he manual and compu e -aided measu emen me hods. Fu he , hal examine s pe o m he manual me hod i s and hen, he compu e ized one. Hal examine s pe o m he compu e ized me hod i s and hen he manual one. Hence, he measu emen p ocess o each OPG is summa ized as ollows: 1) Manual me hod: 30 measu emen s pe o hopan o- mog aphy. Each examine measu es 3 imes each OPG ollowing he manual me hod as desc ibed in Benson BW e al. (11) and Kleme i E & Kolmakow S (14), (Fig. 1). 2) Compu e -aided me hod: 30 measu emen s pe o ho- pan omog aphy. Each examine measu es 3 imes each OPG using he p oposed compu e -aided sys em. Measu emen s ob ained using he manual p ocedu e a e w i en by he examine a he end o he h ee ounds and subsequen ly an ex e nal use in oduces hese al- ues in o a da abase. Measu emen s ob ained using he compu e ized me hod a e s o ed di ec ly in o he da a- base. The ou pu o he sys em is an excel ile o each OPG exhibi ing he di e en measu emen s aken by each examine . S a is ical Analysis: Repea abili y and ep oducibili y a e used o analyze he esul s in e ms o in e - and in a-examine ag ee- men espec i ely. Repea abili y e e s o he a i- abili y o he measu emen s ob ained by one examine (obse e ) while measu ing he same da a epea edly. Rep oducibili y is he a iabili y o he measu emen sys em caused by di e ences in he obse e beha io . These measu es a e es ima ed h ough he p ecision o he measu emen s exp essed as he a iance o mul i- ple measu emen s by a single pe son. Fu he mo e, he ange o eliabili y o he measu emen s is analyzed us- ing he 95% con idence in e al. In addi ion, a e age ime equi ed o es ima e he indexes using bo h me h- ods (manual and compu e -aided) is conside ed. Finally, pai -wise compa isons be ween manual and compu e aided measu emen s a e compu ed by Wilcoxon signi i- cance es (23). Resul s Repea abili y o he measu es is analyzed measu ing he ag eemen achie ed when one use measu es he same in di e en o hopan omog aphs. (Table 1) lis s he di e ences pe use o each index. Fu he mo e, he dis ibu ions o e o s (di e ences) o all he use s a e shown in (Fig. 3A, 3B and 3C), o he h ee indexes espec i ely. As shown, he a iabili y using he p o- posed sys em is lowe han manually compu ing each index. Tha is, di e ences be ween he same use meas- u emen s a e lowe when he examine akes he bene i o he compu e aided sys em. Rep oducibili y o he measu emen s is analyzed using he a e age measu emen s pe use . Gi en an o ho- pan omog aphy, he a e age alue pe use is es ima ed using he di e ences be ween each use measu emen Compu e aided Manual CW PMI M/M CW PMI M/M Use 1 ±0.2922 ±0.0187 ±0.2552 ±0.5724 ±0.0546 ±0.1864 Use 2 ±0.5678 ±00449 ±0.4819 ±0.6806 ±0.0769 ±0.3505 Use 3 ±0.5471 ±0.0431 ±0.3497 ±0.7131 ±0.1136 ±0.5537 Use 4 ±0.4907 ±0.0414 ±0.4777 ±0.8830 ±0.2561 ±0.4900 Use 5 ±0.2530 ±0.235 ±0.3541 ±0.7955 ±0.0911 ±0.4323 Use 6 ±0.2479 ±0.0173 ±0.2240 ±0.7051 ±0.1126 ±0.5016 Use 7 ±0.5304 ±0.0425 ±0.3993 ±0.8165 ±0.0910 ±0.5617 Use 8 ±0.5158 ±0.0408 ±0.4799 ±0.8095 ±0.1179 ±0.5182 Use 9 ±0.3218 ±0.0229 ±0.3180 ±0.8095 ±0.1024 ±0.5531 Use 10 ±0.5499 ±0.0470 ±0.3945 ±0.7665 ±0.1057 ±0.4774 Table 1. Use a iabili y. Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e629 o e he a e age measu emen om all he use s. Each use measu emen is compu ed as he a e age alue o all ials. As a esul , he summa y o alues lis ed in (Table 1, 2A and 2B). Fu he mo e, h ee e o dis ibu ions a e ob ained, one o each index (Fig. 3D, 3E and 3F). The summa y o s a is ics o e hese dis ibu ions and he Wilcoxon signi icance es a e lis ed in (Table 3A). F om hese esul s, we can conclude ha ou compu e -aided sys- em ou pe o ms he manual p ocedu e in e ms o ep oducibili y (23). Time equi ed o compu ing he indexes is analyzed. A e age ime pe analysis is lis ed in (Table 3B). As shown, he a e age ime pe o hopan omog aphy Fig. 3A. Pe use co ical wid h a iabili y. 3B. Pe use pano amic index a iabili y. 3C. Pe use M/M a io a i- abili y. 3D. Dis ibu ion o e o s when di e en examine s measu e he same o hopan omog aphy (OPG). 3E. Dis ibu ion o e o s when di e en examine s measu e he same OPG. 3F. Dis ibu ion o e o s when di e en examine s measu e he same OPG. Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e630 Le Compu e aided Manual CW PMI M/M CW PMI M/M 1 6.663±0.24 0.426±0.01 2.482±0.26 6.067±0.61 0.438±0.24 2.037±0.37 2 4.677±0.19 0.423±0.01 2.714±0.13 5.400±1.31 0.444±0.13 2.287±0.69 3 4.778±0.22 0.387±0.02 1.809±0.04 5.533±0.79 0.379±0.10 2.021±0.36 4 4.653±0.22 0.478±0.02 3.648±0.32 5.300±0.85 0.382±0.09 2.288±0.48 5 4.657±0.18 0.454±0.03 3.267±0.42 4.867±0.53 0.292±0.05 2.049±0.44 6 5.662±0.15 0.400±0.01 1.782±0.10 6.233±0.74 0.319±0.11 1.868±0.35 7 4.972±0.29 0.256±0.01 1.461±0.08 5.200±0.93 0.422±0.08 2.136±0.36 8 6.747±0.19 0.470±0.02 2.487±0.19 7.133±0.71 0.348±0.12 1.774±0.26 9 5.635±0.19 0.365±0.01 2.564±0.11 6.667±0.35 0.307±0.05 1.995±0.38 10 5.415±0.34 0.388±0.01 2.268±0.21 5.100±0.80 0.367±0.62 2.156±0.43 Table 2A. Use measu emen a iabili y o he le side o each o hopan omog aphy. Righ Compu e aided Manual CW PMI M/M CW PMI M/M 1 5.621±0.47 0.393±0.02 1.835±0.21 6.133±0.71 0.367±0.06 2.148±0.40 2 4.716±0.12 0.456±0.01 2.029±0.29 5.867±0.57 0.399±0.15 2.132±0.33 3 2.465±1.38 0.212±0.12 1.343±0.75 5.400±0.64 0.221±0.20 1.119±0.98 4 4.828±0.19 0.529±0.01 2.029±0.29 5.400±0.83 0.393±0.09 2.098±0.43 5 4.735±0.15 0.496±0.04 2.087±0.26 4.333±0.35 0.437±0.09 2.215±0.46 6 5.501±0.24 0.415±0.01 1.263±0.13 5.967±1.24 0.450±0.17 1.754±0.55 7 5.127±0.24 0.284±0.01 1.909±0.19 5.067±0.94 0.382±0.14 1.806±0.30 8 6.557±0.36 0.446±0.02 1.648±0.19 5.800±0.69 0.366±0.07 1.665±0.26 9 5.826±0.28 0.334±0.05 1.766±0.66 6.167±0.91 0.357±0.10 1.781±0.61 10 5.332±0.31 0.333±0.02 1.553±0.26 4.967±0.33 0.330±0.04 1.708±0.55 Table 2B. Use measu emen a iabili y o he igh side o each o hopan omog aphy. index Compu e Aided Manual Wilcoxon es CW ±0.2194 ±0.7605 1 PMI ±0.0174 ±0.1093 1 M/M a io ±0.2076 ±0.4070 1 Table 3A. S anda d de ia ion o e o s when di e en examine s measu e he same o hopan omog aphy. The hi d column lis s he Wilcoxon signi icance es o he ep oducibili y expe imen . A posi i e alue (1) indica es ha he compu e aided sys em ou - pe o ms he manual p ocess. A e age ime S anda d de ia ion Compu e -aided 16,08s. ±7.45s. Manual 156,66s. ±45,95s. Table 3B. A e age ime equi ed o measu e all h ee indexes in a single o hopan omog aphy using he manual and he p o- posed compu e -aided me hods. Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e631 using he compu e aided sys em is 16 seconds. How- e e , he a e age ime equi ed o pe o m he same analysis manually is highe han 2 minu es. Hence, we can conclude han using he compu e aided me hod educes signi ican ly he equi ed ime pe analysis. The e o e, he compu e -aided me hod also educes he use a igue and, consequen ly, imp o es he long e m accu acy o he analysis. Discussion The e a e a numbe o wo ks using pho odensi ome - ic analysis o diagnose os eopo osis. Fo ins ance, Na- kamo o T e al. (24) in 2008, pe o m g ay le el analysis in OPG o iden i y pa ien s wi h os eopo osis. Jonasson G e al. (25) in 2001, p edic he skele al DMO using a pho odensi ome ic measu emen o he bone mass o he mandible, he abecula pa e n o he al eola bone and i s in e den al hickness. O he au ho s de ine a c i- e ion o classi y abecula pa e ns as dense, he e oge- neous o spa se (26,27). They conclude ha he isual assessmen o he abecula pa e n in OPG can be used o iden i y pa ien s ha ing os eopenia o os eopo osis. highly in luenced by he expe ise o examine s. These au ho s compa e manual and compu e ized me hods o es ima e he co ical wid h. As a esul , hey conclude ha bo h sys ems a e highly co ela ed al hough hey do no speci y he sys em used. Howe e , esul s ega ding ime o deg ee o di icul y a e no p o ided (19). The p oposed compu e -aided sys em is use iendly and does no equi e wo much ime o es ima e he meas- u es. Fu he , gi en an o opan omog aph, i is highly epe i i e o a gi en examine . In addi ion, gi en he low examine a igue p oduced, in aexamine calib a- ion using he epea abili y index p oposed by Landis JR & Koch GG (28) is easible. Fu he mo e, he e a e no e o s associa ed wi h he anno a ion p ocess since i is done au oma ically. Tha is, he sys em p o ides an excel ile as ou pu wi h all he measu emen s pe o opan omog aph. Finally, he p ecision o he compu- e -aided me hod in he measu emen s in on o he manual one mus be highligh ed. Example measu es o a single o hopan omog aphy a e lis ed in (Table 4). As shown, measu ing he co ical wid h manually leads o ough measu es. Howe e , using he compu e –aided Compu e aided Manual Righ Le Righ Le CW PMI M/M CW PMI M/M CW PMI M/M CW PMI M/M Use 1 4,97 0.37 2,07 6,14 0,40 2,27 5 0.33 1.94 5 0.42 2.05 Use 2 4,23 0.32 2,56 6,69 0,41 2,34 4 0.27 1.45 6 0.32 1.40 Use 3 5,76 0.40 2,02 6,41 0,42 2,39 5 0.34 2.34 6 0.43 2.59 Use 4 5,36 0.37 1,78 6,42 0,42 2,25 5 0.36 2.80 7 0.23 2.24 Use 5 6,11 0.41 1,31 6,80 0,43 2,25 5 0.43 0.90 5 0.27 0.98 Use 6 5,71 0.39 2,32 6,60 0,42 2,36 5 0.42 1.80 5 0.30 1.89 Use 7 5,84 0.40 1,61 6,68 0,41 2,63 6 0.43 0.40 4 0.17 2.34 Use 8 5,61 0.38 1,34 6,41 0,38 2,25 5 0.33 1.70 4 0.26 2.65 Use 9 6,32 0.42 1,28 7,62 0,47 2,29 5 0.28 1.10 4 0.33 1.29 Use 10 6,18 0.42 2,63 6,40 0,41 2,46 5 0.51 0.80 4 0.21 2.30 Table 4. Example measu es o a gi en o hopan omog aphy using bo h me hods: compu e -aided and manual. Rega ding manual measu emen s in den al pano amic adiog aphs, he e a e only a ew wo ks add essing hem using a compu e ized sys em. A i in AZ e al. (19) in 2006, use a compu e -aided me hod in digi al adio- g aphs o es ima e MTC. The sys em elies on image enhancemen echniques o measu e MTC. Howe e , since he ma gins o he co ex a e no always clea ly de- ined he sys em equi es manual assis ance o pe o m i s ask. Fu he , he a iabili y o he measu emen s is sys em p o ides highe accu acy educing he p opaga- ion o e o s. On he con a y, he e a e di e en di - icul ies associa ed o he manual measu emen me hod such as speci ic illumina ion, magni ica ion, examine calib a ion as p oposed in Ledge od e al. (29) o sys- ema ic measu emen s as p oposed by Ka ayianni e al. (30). Fu he , he manual p ocedu e is ulne able o e - o s when anno a ing egis e ed measu es. Med O al Pa ol O al Ci Bucal. 2012 Jul 1;17 (4):e624-32. Compu e -aided measu emen s in den al pano amic adiog aphs e632 Conclusions We ha e p oposed a e y use iendly compu e ized me hod o measu e h ee di e en mo phome ic man- dibula indexes: co ical wid h, pano amic mandibula index and mandibula al eola bone eso p ion index. The p oposed me hod exhibi s supe io epea abili y and ep oducibili y a es compa ed o s anda d manual me hods. Mo eo e , he ime equi ed o pe o m he measu emen s using he p oposed me hod is negligible compa ed o pe o m he measu emen s manually. F om he esul s we can conclude ha he sys em p o ides a p ac ical manne o pe o m hese measu emen s. I does no equi e an expe examine and does no ake mo e han 16 seconds pe analysis. Thus, i may be sui - able o diagnose os eopo osis using den al pano amic adiog aphs. Re e ences 1. Na as Cáma a FJ, Fe nández de San iago FJ, Bayona Ma zo I, Mingo Gómez T, de la Fuen e Sanz MM, Cacho del Amo A. P e a- lence o os eopo osis assessed byquan i a i e ul asound calcaneus measu emen s in ins i u ionalized elde ly popula ion. An Med In e - na. 2006;23:374-8. 2. Fi e J, Nolla JM, Gómez-Vaque o C, Ma ínez-Aguilá D, Val e - de J, Roig-Esco e D. A compa a i e s udy o compu ed digi al ab- so p iome y and con en ional dual-ene gy X- ay abso p iome y in pos menopausal women. Os eopo os In . 2001;12:565-9. 3. Na es Díaz M. Os eopo o ic ac u es in Spanish women. Med Clin (Ba c). 2006;127:413-4. 4. Comps on JE, Papapoulos SE, Blancha d F. Repo on os eopo- osis in he Eu opean Communi y: cu en s a us and ecommenda- ions o he u u e. Wo king Pa y om Eu opean Union Membe S a es. Os eopo os In . 1998;8:531-4. 5. Roig Vilaseca D, Gómez Vaque o C. Wha clinical guideline should we use o he managemen o os eopo osis?. Med Clin (Ba c). 2011;136:293-4. 6. Wo ld Heal h O ganiza ion. Assessmen o ac u e isk and i s applica ion o sc eening o pos menopausal os eopo osis. Repo o a WHO S udy G oup. Wo ld Heal h O gan Tech Rep Se . 1994;843:1- 129. 7. Ho ne K, De lin H. The ela ionship be ween mandibula bone mine al densi y and pano amic adiog aphic measu emen s. J Den . 1998;26:337-43. 8. Whi e SC. O al adiog aphic p edic o s o Os eopo osis. De ma o- maxillo ac Radiol. 2002;31:84-92. 9. Kaye EK. Bone heal h and o al heal h. J Am Den Assoc. 2007;138:616-9. 10. Taguchi A, Tanimo o K, Suei Y, O ani K, Wada T. O al signs as indica o s o possible os eopo osis in elde ly women. O al Su g O al Med O al Pa hol O al Radiol Endod. 1995;80:612-6. 11. Benson BW, P ihoda TJ, Glass BJ. Va ia ions in adul co ical bone mass as measu ed by a pano amic mandibula index. O al Su g O al Med O al Pa hol. 1991;71:349-56. 12. Yasa L, Sönmez AS, U ku N, Ozcan J, Cebi Z, Sa an K, e al. E ec o misop os ol on bone mine al densi y in women wi h pos menopausal os eopo osis. P os aglandins O he Lipid Media . 2006;79:199-205. 13. B as J, an Ooij CP, Ab aham-Inpijn L, Kusen GJ, Wilmink JM. Radiog aphic in e p e a ion o he mandibula angula co ex: A diagnos ic ool in me abolic bone loss. Pa I. No mal s a e. O al Su g O al Med O al Pa hol. 1982;53:541-5. 14. Kleme i E, Kolmakow S. Mo phology o he mandibula co ex on pano amic adiog aphs as an indica o o bone quali y. Den o- maxillo ac Radiol. 1997;26:22-5. 15. Taguchi A, Asano A, Oh suka M, Nakamo o T, Suei Y, Tsuda M, e al. OSPD In e na ional Collabo a i e G oup. Obse e pe o - mance in diagnosing os eopo osis by den al pano amic adiog aphs: esul s om he os eopo osis sc eening p ojec in den is y (OSPD). Bone. 2008;43:209-13. 16. Caku B, Dagis an S, Sahin A, Ha o li A, Yilmaz A. Reliabili y o mandibula co ical index and mandibula bone mine al densi y in he de ec ion o os eopo o ic women. Den omaxillo ac Radiol. 2009;38:255-61. 17. De lin H, Ho ne K. Mandibula adiomo phome ic indices in he diagnosis o educed skele al bone mine al densi y. Os eopo os In . 2002;13:373-8. 18. Whi e SC, Taguchi A, Kao D, Wu S, Se ice SK, Yoon D, e al. Clinical and pano amic p edic o s o emu bone mine al densi y. Os eopo os In . 2005;16:339-46. 19. A i in AZ, Asano A, Taguchi A, Nakamo o T, Oh suka M, Tsuda M, e al. Compu e -aided sys em o measu ing he mandibula co - ical wid h on den al pano amic adiog aphs in iden i ying pos me- nopausal women wi h low bone mine al densi y. Os eopo os In . 2006;17:753-9. 20. Taguchi A, Tsuda M, Oh suka M, Kodama I, Sanada M, Naka- mo o T, e al. Use o den al pano amic adiog aphs in iden i ying younge pos menopausal women wi h os eopo osis. Os eopo os In . 2006;17:387-94. 21. Taguchi A, Oh suka M, Tsuda M, Nakamo o T, Kodama I, In- agaki K, e al. Risk o e eb al os eopo osis in pos -menopausal women wi h al e a ions o he mandible. Den omaxillo ac Radiol. 2007;36:143-8. 22. Vlasiadis KZ, Damilakis J, Veleg akis GA, Skou e is CA, F a- gouli I, Goumenou A, e al. Rela ionship be ween BMD, den al pano- amic adiog aphic indings and biochemical ma ke s o bone u no- e in diagnosis o os eopo osis. Ma u i as. 2008;20:226-33. 23. Wang L, Li R. Weigh ed Wilcoxon- ype smoo hly clipped abso- lu e de ia ion me hod. Biome ics. 2009;65:564-71. 24. Nakamo o T, Taguchi A, Oh suka M, Suei Y, Fuji a M, Tsuda M, e al. A compu e -aided diagnosis sys em o sc een o os eopo- osis using den al pano amic adiog aphs. Den omaxillo ac Radiol. 2008;37:274–81. 25. Jonasson G, Bank all G, Kilia idis S. Es ima ion o skele al bone mine al densi y by means o he abecula pa e n o he al eola bone, i s in e den al hickness, and he bone mass o he mandible. O al Su g O al Med O al Pa hol O al Radiol Endod. 2001;92:346- 52. 26. Lindh C, Ho ne K, Jonasson G, Olsson P, Rohlin M, Jacobs R, e al. The use o isual assessmen o den al adiog aphs o iden i ying women a isk o ha ing os eopo osis: he OSTEODENT p ojec . O al Su g O al Med O al Pa hol O al Radiol Endod. 2008;106:285- 93. 27. De lin H, Ka ayianni K, Mi sea A, Jacobs R, Lindh C, an de S el P, e al. Diagnosing os eopo osis by using den al pano amic a- diog aphs: he OSTEODENT p ojec . O al Su g O al Med O al Pa- hol O al Radiol Endod. 2007;104:821-8. 28. Landis JR, Koch GG. An applica ion o hie a chical kappa- ype s a is ics in he assessmen o majo i y ag eemen among mul iple obse e s. Biome ics. 1977;33:363-74. 29. Ledge on D, Ho ne K, De lin H, Wo hing on H. Radiomo - phome ic indices o he mandible in a B i ish emale popula ion. Den omaxillo ac Radiol. 1999;28:173-81. 30. Ka ayianni K, Ho ne K, Mi sea A, Be kas L, Mas o is M, Jaco- bs R, e al. Accu acy in os eopo osis diagnosis o a combina ion o mandibula co ical wid h measu emen on den al pano amic adio- g aphs and a clinical isk index (OSIRIS): he OSTEODENT p ojec . Bone. 2007; 40:223-9.