Application of High-Resolution 3D Scanning in Medical Volumetry

INTL JOURNAL OF ELECTRONICS AND TELECOMMUNICATIONS, 2016, VOL. 62, NO. 1, PP. 23–31 Manusc ip ecei ed No embe 15, 2015; e ised Ma ch, 2016. DOI: 10.1515/ele el-2016-0003 Applica ion o High-Resolu ion 3D Scanning in Medical Volume y Adam Ch omy Abs ac —This pape deals wi h applica ion o 3D scanning echnology in medicine. Impo an p ope ies o 3D scanne s a e discussed wi h emphasize on medical applica ions. Cons uc ion o medical 3D scanne acco ding o hese speci ica ions is de- sc ibed and p ac ical applica ion o i s use in medical olume y is p esen ed. Besides olume y, such 3D scanne is usable o many o he pu poses, like moni o ing o eco e y p ocess, e gonomic splin manu ac u ing o in lamma ion de ec ion. 3D scanning in oduces no el olume ic me hod, which is compa ed wi h s anda d me hods. The new me hod is mo e accu a e compa ed o p esen ones. P inciples o his me hod a e discussed in pape and i s accu acy is e alua ed and expe - imen ally e i ied. Keywo ds—medical olume y, medical 3D scanning, 3D scan- ning, medical imaging, so - issues I. INTRODUCTION Cap u ing h ee-dimensional models has become mo e and mo e impo an du ing las yea s, due o apid de elopmen o 3D echnologies, especially 3D p in e s. They a e e y e- quen ly eques ed de ices oday, i s ma ke is apidly g owing and i s de elopmen also mo es o wa d e y as . Acco ding o Google T ends, numbe o sea ch que ies ela ed o 3D echnologie inc eased mo e han en imes du ing las h ee yea s [1]. A chi ec u al models o buildings, design p o o ypes o new p oduc s, o e en 3D p in e s i sel a e 3D p in ed oday. The e a e expe imen s wi h p in ing eal houses, ood, o biop in ing [2]. All hese applica ions equi e he same – compu e 3D model. 3D p in ing is no he only domain, whe e 3D models a e use ul. They can be also used o s o ing isual in o ma ion in compac and esis an o m, in which he objec s a e no ageing. In his case, colou -co e ed 3D models seem o be he bes modali y. E.g. The Me opoli an Museum o A published models o i s exhibi s [3], o Sou h Ko ea a chi ed i s UNESCO he i age a eas [4]. Finally, compu e 3D models a e, due o i s plas ici y, becoming mo e and mo e used o isualiza ion o objec s, which a e un eachable (con amina ed, dange ous o emo e a eas) [5], [6] o en i onmen s, which a e in isible wi hou in asi e su ge y (human inne s uc u es) [7]. Ano he objec s a e isible, bu i s impo an de ails a e oo iny (human ou e s uc u es) and i is necessa y o enla ge hem plas ically [8]. This wo k was suppo ed by he g an No. FEKT-S-14-2429 The esea ch o new con ol me hods, measu emen p ocedu es and in elligen ins umen s in au oma ion inanced om in e nal science und o B no Uni e si y o Technology. Au ho is wi h Facul y o Elec ical Enginee ing and Communica- ion, B no Uni e si y o Technology, B no, Czech Republic (e-mail: [email p o ec ed].cz). The e a e se e al possibili ies o cap u ing such 3D models, bu he mos equen me hod is 3D scanning, wha means di ec cap u ing o 3D model wi h de ice in ended o his pu - pose [9]. This echnology is sp eading in o he new domains and mo e and mo e new applica ions a e announced each yea . Bu s ill, he e a e mos ly echnical domains ge ing in ol ed. The medicine is one o domains, whe e 3D models a e e y a ely used, e en hough many oppo uni ies o po en ial 3D scanning applica ions exis . This pape deals wi h applica ions o 3D scanning ech- nologies in medicine, wi h ocus on one o he a eas, whe e 3D scanning b ings signi ican ad an ages – on Medical Volume y. Obse ing pa ame e s o human body olumes has been one o he mos impo an ac o s in diagnosis since beginnings o medicine and has also se ed o e alua ion o sui abili y o applied he apy. Bu i you a e isually obse ing only, all hese geome ical changes can be use ul once he symp oms a e la ge, wha ypically means ha he disease is al eady de eloped. To be able o e eal he disease in i s e y beginning, i is no enough o obse e only, i needs o measu e and e en wi h as high as possible esolu ion, wha p o ides possibili y o ea ly de ec ion o iny changes o olumes, which e e o symp oms o disease. In his s udy, we a e ocused on applica ion o p ecise 3D scanning o accu a e measu emen s o olume ic pa ame e s, p ima ily body-pa olume. To know he exac alue o his pa ame e is aluable o many pu poses – e.g. ea ly de ec ion o pe iphe al oedemas [10], lymphedemas, ca cinomas [11] o ib osis [12], i s moni o ing and con ol o i s e olu ion; measu emen o in luence o s eng h exe cises on spo smen [13] o supe ision o eco e y p ocess a e in asi e su ge ies [14]. P esen s anda d me hods a e no accu a e enough, easily usable, o equi e high ope a ional expenses. P ecise 3D scanning has also one ex a ad an age – spa ial esolu ion o such models can be also enla ged by one dimen- sion and all hese models could be egis e ed in ela ion o ime. Wi h 3D scanning, we a e no able o p ecisely measu e only, we a e also able o see ends. In his pape , he p esen olume ic me hods a e summa- ized, hen cons i u ion o he bes sui able 3D scanne o medical pu poses is discussed and i s p ac ical applica ion in medical olume y is p esen ed. Finally, his new olume ic me hod is compa ed wi h p esen ones, bo h heo e ically and expe imen ally. 24 A. CHROMY II. PRESENT VOLUMETRIC METHODS Acco ding o [11], [13], [15]–[17], he mos equen ly used olume ic me hods a e: A. Ci cum e en ial measu emen s Me hods o his g oup a e based on olume es ima ion om measu emen s o ci cum e ence a se e al speci ic places. E e y me hod uses some o m o su ace app oxima ion, wha leads o lowe accu acy [16]. Repea abili y o measu emen s signi ican ly depends on expe ience o pe son pe o ming measu emen [18]. On he o he hand, no special equipmen is equi ed, hey a e simple and use ul o non- lexible limbs and pa ien s wi h bad mo o ic abili ies o wa e incompa ible diseases [11]. The i s mos used me hod is F us um Sign Model based on measu emen s o 2 ci cum e ences and app oxima ion by unca ed cone be ween hem, wi h ela i e accu acy1abou ±8% o measu ed olume [18]. E en hough i s accu acy is low, i s ex emely as (less han 1 min.), wha makes i sui able o si ua ions, whe e quick es ima ion o olume, wi hou emphasis on accu acy, is equi ed [19]. Second me hod called Disc Model es ima es olume as sum o equidis an disks wi h dis ance o 5 cm, wi h ela i e accu acy abou ±6% o measu ed olume [18], bu be awa e, ha all hese accu acies mus no been conside ed as de ini i e since signi ican ly depends on pe sonal expe iences. B. Wa e Displacemen Volume y The mos used olume ic me hod, conside ed as golden s anda d [13], is based on quan um o wa e o e lowing om ully illed con aine when measu ed limb is inse ed. I is equen ly used because o i s good accu acy, e y good epea abili y and negligible dependency on ope a o ’s expe iences compa ing o ci cum e en ial measu emen s. The bigges disad an ages a e, ha i equi es good lexibili y and good mo o ic abili ies o measu ed limb (shi e ing o limb signi ican ly in luences esul ), he e is possible isk o c oss- in ec ion, some pa ien s ha e o a oid wa e because o hei disease and i is e y ime consuming [11], [15]. Acco ding o [13], [18], [20], i s ela i e accu acy is in e sely p opo ional wi h measu ed objec olume and a ies om ±2% in case o lowe leg measu emen s (abou 2700 cm3) o ±8% in case o inge measu emen s (abou 25 cm3). C. Op oelec onic Volume Measu emen The e a e se e al comme cially a ailable single-pu pose olume ic de ices using his me hod. The measu emen p in- ciple is based on ho izon ally mo able ame equipped wi h in a ed ligh emi e s and ecep o s. This ame is mo ing along axis o examined limb. Ligh beams a e in e up ed by he olume o measu ed limb and he shadow is cap u ed a ecep o s in wo pe pendicula axes. By mo ing he ame, many measu emen s a e pe o med and olume is hen cal- cula ed om hese alues [18]. Acco ding o [20] ela i e 1 e m ” ela i e accu acy” in his pape means ela i e unce ain y wi h 95% con idence (2σ) accu acy o such de ice is ±2% o measu ed olume (in case o measu emen s in ange om 1000 o 3000 cm3), bu he epea abili y is much be e since he measu ed alue is no oo dependan on pe sonal expe iences o mo o ic abili ies o pa ien [16]. Disad an age o his me hod is, ha expensi e, single-pu pose de ice is equi ed and esul an bene i s a e no oo signi ican compa e o Wa e Displacemen Volume y, which is much cheape [10]. D. 3D Sonog aphy, CT and MRI In e y occasional cases, mos ly o esea ch pu poses, he olume is compu ed om 3D models p o ided by hese h ee- dimensional imaging modali ies [21]–[24]. Each o hem is able o p o ide 3D model o body pa , bu each o hem has signi ican disad an ages disallowing i s usage in common heal h ca e. Ul asonog aphy, due o i s di e gence o sound- beam, has oo poo esolu ion (abou 5 mm) o be be e han simple Ci cum e en ial o Wa e displacemen me hods [25]. Among his, i has oo noisy image [26]. Compu ed Tomog aphy (CT) eaches up o 0.2 mm spa ial esolu ion in ou pu 3D model [27], bu ionizing adia ion abso bed by pa ien du ing one scan is up o 15 mS , wha is one hi d o allowed exposi ion o wo ke s wi h ionizing sou ce pe yea and o common people exceeds allowed hygiene limi s e en 15 imes [28]. Fo his eason, use o his modali y is allowed as a e as possible and epea ed scanning is comple ely ou o he ques ion. Finally, Magne ic Resonance Imaging (MRI), al hough i s esolu ion is app ox. 1 mm, canno be widely used because o i s high cos s, ime consuming examina ions, pacemake o pie cing disallowing [29]. Applica ion o 3D scanning in medicine b ings no el olume - ic me hod, which eaches o be e accu acy han men ioned olume ic me hods, good epea abili y, easy use and low ope a ional cos s. III. 3D SCANNNERS The e a e plen y o 3D scanne s wo king on a ious p in- ciples, so hei e iew would ake a whole book. Bu wha hey ha e common is, ha we can make a lis o modules, which each scanne mus ha e somehow implemen ed. Ra he hen e alua ing each model o 3D scanne , we discuss each a ailable implemen a ion o e e y module wi h emphasize on medical applica ion o 3D scanne . By choosing he mos sui able implemen a ion o each unc ional block, we can make an image, how he ideal medical 3D scanne should be ealized. The e a e blocks, which each scanne consis s o : A. Raw Da a Cap u ing The mos o scanne s measu e dis ance o su ace om senso . In case o con ac scanne s, his dis ance is gi en by leng h o ouch p obe. This can be e y p ecise, bu ouching he examined objec can be damaging o us a ing (in case o human body scanning) [30]. Con ac less echniques a e mos ly using he lase beam (lase scanne s). Solu ion wi h measu ing he ime o ligh o APPLICATION OF HIGH-RESOLUTION 3D SCANNING IN MEDICAL VOLUMETRY 25 lase quan um p o ides wide measu ing ange (up o hund eds o me e s), bu i s p ecision is limi ed by esolu ion o ime- measu ing uni (up o 1mm) [31]. Ano he lase scanne p inciple is iangula ion, ha ing opposi e abili ies: i s accu- acy is e y high (up o 1µm [32]), bu i s measu ing ange is limi ed (hund eds o millime es [33]). The in e e ome ic lase scanne eaches up o 1pm esolu ion, bu i s ange is up o hund eds o µm [34]. S uc u ed-ligh 3D scanne s p ojec a pa e n o ligh on he subjec and look a i s de o ma ion in cap u ed image [35]. The ad an age o his me hod is high speed o scanning since each ime he en i e image is analysed ins ead o one poin [36]. Ano he ad an age is he p ecision, which is e en be e han lase iangula ion [37]. S e eopho og amme ic scanne s use wo came as o es- ima e 3D posi ion o poin s om di e ences be ween wo images. I is also qui e as wi h accu acy up o 1mm [38]. Di e en ype o aw da a cap u e MRI and CT, whe e he alue ele an o densi y o ma e ial, is measu ed in e e y place o examined egion. MRI measu emen s a e e y slow ( ens o minu es), bu inne s uc u es o human body a e isualized (wi h spa ial esolu ion app ox. 1.5mm3) [21], [22]. Sligh ly as e and mo e accu a e (app ox. 0.6mm3) is CT [27], bu i s impo an disad an age is ionizing adia ion. B. Mo ing he Senso In o de o build a 3D model o a bi a ily complex objec , he senso mus be posi ioned o se e al iew-poin s, om which all de ails on he su ace o objec a e isible [39]. The hand-held scanne s a e de ices, whe e mo ing o senso along scanning ajec o y is ealised manually [40]. I eaches he bes lexibili y, bu he ask o senso localiza ion (see nex sec ion) is complica ed [41]. The majo i y o 3D scanne s use mo o ized senso mo ing, mos ly composed om p ecise elec ic s eppe d i e [42] o se omo o [43]. I p o ides au oma ic mo emen s along op i- mal, p e- es ed ajec o ies and empowe s also easie senso localiza ion. I also minimizes he p oblem wi h scanning anges, which can be e y na ow in some cases (e.g. iangu- la ion lase scanne s) and o s ay manually in his ange can be di icul . Disad an age is, ha lexibili y o mo emen is limi ed by he kinema ic concep ion o axes, along which he senso mo es [44]. The bes lexibili y eaches he a icula ed kinema ic chain wi h a leas 6 DOF. When adding 7 h deg ee o eedom, he singula i y p oblems a e minimized [45]. C. Senso Localiza ion To be able o esol e posi ion o measu ed poin in coo - dina e sys em, i is necessa y o know he p ecise posi ion o senso in 6 DOF2(among aw da a i sel ). This ask is simple in cases, when mo o ized senso posi- ioning is used, since he senso mus be connec ed o mo e- able axis by join s3, which posi ion can be simply measu ed 23 coo dina es unambiguously de ine posi ion o poin and ano he 3 coo dina es unambiguously de ine i s o ien a ion 3Join in his sense can be also linea ly ansla ional, no only o a ional by encode s, esol e s [46], se omo o ic eed-back [43], e c. Using di ec kinema ics, i s posi ion can be unambiguously compu ed wi h high p ecision [47]. F ee hand-held scanne s use he image egis a ion o es- ima e i s loca ion al e a ion om change o scene be ween las wo scans [48]. Many scanne s use some me hod based on ICP [49], usually op imized [50] o modi ied o speci ic applica ion [51]–[53]. This app oach has wo disad an ages: The absolu e accu acy is e y low, due o cumula i e cha ac e o localizing algo i hm, which sums pa ial e o s [54]. I can also ail in uni o m scenes like a plane, cylinde , sphe e, e c. This p oblem can be pa ly minimized by ex e nal localizing sys em based on ine ial measu emen uni (IMU) [41], o ully sol ed by use o passi e kinema ic chain wi h join measu emen s [55]. D. Planning T ajec o y o Senso The e a e 3 ypes o scanning ajec o y planning: online, s a ic and adap i e. Online planning is ypical o hand-held scanne s, since he exac ajec o y is c ea ed jus when scanning and is unknown be o e [40]. The esul depends on ope a o skills (i he mee s he measu ing ange, i scanning jus he egion o in e es , e c.). The o he special example o online planning is au onomous planning o ajec o y based on ex a sys em measu emen s (e.g. o e iew lase scanne ) [56]. E en hough many publica ion ha e been published on his opic, he solu ion a e s ill no enough obus o be commonly used [57]. S a ic planning uses p ede ined and p e- es ed scanning ajec o ies. They a e simple and obus in sense o no isk o unexpec ed beha iou o posi ioning sys em [58]. On he o he hand, hei use is non- lexible and limi ed e.g. wi h senso s wi h na ow measu ing ange. T ajec o ies mus be equen ly ede ined also in case o sligh di e ences o scanned objec s. Adap i e planning combines bo h me hods abo e. The gen- e al concep o ajec o y is s ic ly de ined, bu sligh , on-line compu ed de ia ions acco ding o scanned objec a e allowed (Fig. 1) [58]. Since he changes o ajec o y a e allowed jus in de ined deg ee o eedom and also limi ed, he e is no isk o unexpec ed mo ing and he algo i hm is mo e lexible. E. Compu a ion o 3D Poin Posi ion The con ac scanne s dispose wi h he easies way o 3D poin compu a ion, since he posi ion o ouched poin is di ec ly he senso loca ion, only shi ed by he leng h o p obe [30]. Fig. 1. Illus a ing di e ence be ween S a ic planning and Adap i e planning. 26 A. CHROMY The mos o 3D scanne ypes use geome ic ans o ma- ions o build he 3D model. This echnique can be used, when cap u ed aw da a con ains any spa ial in o ma ion (e.g. dis ance measu ed by lase scanne ) o a e spa ially o de ed ( oxels o CT and MRI). In case o s e eopho og amme y, he ans o ma ion ap- p oach canno be used, since he e is no spa ial in o ma ion in he aw cap u ed da a. The h ee-dimensional ep esen a ion o scanned objec mus be in e ed om wo images cap u ed wi h wo came as wa ching he same scene om di e en iews [48]. F. S o ing and Visualiza ion o 3D Models The e a e wo basic app oaches o s o ing h ee-dimensional da a: 3D G id di ides spa ial a ea o scanned objec o egula ly o de ed same-size elemen s called oxels. Each cube con ains p obabili y o occupancy by objec (Occupancy G id [59]) o o he alue desc ibing he oxel locali y (E idence G id [60]), e.g. densi y in case o MRI o CT. All objec s in scan a e hen app oxima ed by numbe o same cubes, wha is a om eali y. Also he memo y equi emen s a e e y high. On he o he hand, compu a ions o e such model a e simple and as . I s use ul o models in low esolu ion, which a e equen ly upda ed and ebuild, e.q. in case o 3D li e iew. This ep esen a ion is used in s anda d medical ile o ma DICOM, used o s o ing da a om CT and MRI. Poin Cloud in i s aw o ma is uno de ed se o h ee- dimensional ec o s de ining posi ion o poin s in space [60]. In p ac ical applica ions, hey con ains also ex a in o ma ions abou his poin (colou , ma e ial, e c.). This o ma allows o desc ibe he wo ld by geome ic shapes, wha is mo e au hen ic app oxima ion han g ids. The memo y equi emen s a e signi ican ly lowe han g ids. Disad an age o his o ma is highe compu a ional load and mo e complex analysing o model. I s use ul o models, whe e p ecision and high esolu ion is equi ed, bu hey a e no upda ed oo o en [61]. IV. MEDICAL 3D SCANNER The uni e sal 3D scanning de ice sui able o medical pu poses and usable in e e yday p ac ice shall ha e ollowing abili ies: •High accu acy – essen ial pa ame e o being able o dis inguish e en iny changes o human body caused by oedemas, muscle a ophy o muscle s eng hening. •Flexibili y – since de ice should be uni e sal, we shall be capable o scanning en i e body as same as i s iny de ails. Because o ha , he 3D scanne mus be e y lexible. •Low ope a ional cos s – o allow i s e e yday use, i s ope a ion shall be inexpensi e. •Simple manipula ion – de ice mus be as much as possi- ble au oma ed, no dis u bing he pe sonnel wi h complex se ings be o e each scanning. •High speed – he scanning p ocedu e mus be e y as . In o he cases, he pe sonnel would no ha e ime o use i and would p e e es ima ion ins ead o measu emen . •No limi a ions – de ice should be usable wi h any pa ien . The e should be no limi a ions acco ding o me al pa s, heal h s a e, e c. •Ha mless ope a ion – using he de ice shall no be ha m- ul o bo h pa ien and pe sonnel in any ci cums ances. To ul il all hese equi emen s, he design o medical 3D scanne shall be as ollows: Da a cap u ing senso shall be s uc u ed ligh 3D scanne o iangula ion lase scanne , due o p ecision easons. As a esul o his, compu a ion o 3D poin posi ion shall be based on geome ic ans o ma ions. Senso mo ion shall be mo- o ized om eason o p ecise senso localiza ion, au oma ic mo emen and keeping in measu ing ange. The kinema ic chain shall be a icula ed, due o he lexibili y equi emen . T ajec o y planning shall be adap i e, in o de o each some deg ee o au onomy wi h keeping he mo ion unde he con ol. Cap u ed da a shall be s o ed in o m o ex ended poin cloud, since i does no a ec model accu acy. The p oblem wi h compu a ional equi emen s is no se ious, since he model is once cap u ed and a e wa ds modi ied occasionally. Because such de ice is no comme cially a ailable, we c ea ed ou own 3D scanne [62] mee ing he speci ica ions abo e: Robo ic 3D Scanne . A. Robo ic 3D Scanne Robo ic 3D scanne is a 3D modelling sys em based on no el cons i u ion, which uses combina ion o 6 DOF indus ial obo ic manipula o and iangula ion-based lase scanne connec ed wi h con olling and da a p ocessing com- pu e (Fig. 2). Lase scanne measu es dis ance om pa ien su ace, when obo ic manipula o con ols scanne posi ion and o ien a ion. F om in o ma ion abou posi ion, di ec ion o iew and dis ance o su ace, each poin posi ion in 3D space can be compu ed. This solu ion combines high lexibili y wi h high p ecision and eliabili y. Fig. 2. Robo ic manipula o wi h lase scanne combines high accu acy wi h lexibili y. APPLICATION OF HIGH-RESOLUTION 3D SCANNING IN MEDICAL VOLUMETRY 27 Fig. 3. Analyzing so wa e allowing measu emen s o egion o in e es . High accu acy is eached by using p ecise manipula o wi h accu a e lase scanne and high lexibili y is caused by p o- g ammable scanning ajec o y in six deg ees o eedom and by eplaceabili y o lase scanne , wha p o ides possibili y o scanning bo h iny and la ge s uc u es. I is 3D modelling sys em, use ul o many di e en medical applica ions beside olume ics: moni o ing o issue eco e y p ocess, pose measu emen s o ehabili a ion pu poses o e gonomic splin s design; bu also a o he domains om a chi ing o his o ical ma e ials in museums, h ough design, models o compu e games and indus ial inspec ion o 3D objec cloning. Robo ic 3D Scanne p o ides h ee-dimensional model o pa ien body-pa , which can be analyzed in ou sel -de eloped analyzing so wa e (Fig. 3). The big ad an age is he pos- sibili y o selec ion o egion o in e es . The e a e se e al pa ame e s a ailable o be measu ed: •Dis ances [mm] – be ween de ined poin s (di ec ly, along he su ace), ci cum e ences o ROI. •Angles [deg] – angle be ween h ee de ined poin s (e.g. e eb ae posi ions) •Su ace a ea [mm2] – en i e model o ROI •Volume [mm2] – en i e model o ROI de ined by cu ing plane o de lec ed cu ing su ace De ailed desc ip ion o Robo ic scanne ’s p inciples and unc ions, ele an ans o ma ion equa ions o analysing so - wa e capabili ies can be ound a [44] o [8]. V. NEW VOLUMETRIC METHOD Con a y o F us um Sign Model, Disc Model, Wa e Dis- placemen Me hod and Op oelec onic Volume e s, wha a e he single-pu pose me hods o de ices, his me hod applies mul i-pu pose Robo ic 3D Scanne [44] in o de o measu e olume. Likewise he MRI o CT, i is no a de ice de eloped o measu e olumes only, i is a uni e sal 3D scanning de ice applicable o many o he a ious pu poses. Robo ic 3D Scanne Volume y use he same measu ing p o- cedu e as olume y ealized by MRI, CT o Ul asonog aphic olume y. The same measu ing p ocedu e is applied also in case o using any o he gene ic 3D scanne , he only di e ence a e p ope ies o ou pu alues (accu acy, epea abili y, egion selec abili y, e c.). Fig. 4. Main p inciple o p oposed olume ic me hod. Me hod p inciple can be di ided o se e al s eps, as shown on Fig. 4. Fi s s ep is cha ac e ized by c ea ing a p ecise 3D model o measu ed limb o any o he in e es ed body pa . This model is s o ed in poin -cloud o m, which elimina es losses o de ailed ea u es and inc eases accu acy o he me hod 4. Such models a e isualized o ope a o , who de ines egions o in e es (ROI) whe eo olume wan o be measu ed (Fig. 3). This abili y is an impo an ad an age on con a y o some s anda d me hods (e.g. Wa e Displacemen Volume y), whe e ROI is de ined by me hod i sel . The olume o selec ed ROI is hen compu ed. Robo ic 3D Scanne Volume y uses Signed Volume o Te ahed on Me hod [63], which accu acy is limi ed only by posi ioning e o o poin s o cap u ed 3D model. G id-based s o ing sys ems (MRI, CT, Sono) compu es olume as sum o oxels5 belonging o ROI. Resul ing accu acy is hen limi ed by g id esolu ion. A. Majo Ad an ages o This Me hod The e a e se e al majo ad an ages o his me hod, com- pa ed o s anda d ones: •The highes accu acy compa e o s anda d me hods (see sec ion V-B). •Accu acy is independen on skills o hospi al s a , like i is no in case o Ci cum e en ial measu emen s. •Fas measu ing p ocess compa ed o all he me hods excep he Op oelec onic olume e s. •No equi emen s on pa ien like a Wa e Displacemen Me hod (mobili y and lexibili y o limb) o MRI (no pie cing, no pace-make s). •Mul i-pu pose imaging modali y, no single-pu pose de- ice as Op oelec onic olume e s. •Selec i i y o measu ed a ea (possible o de ine egion o in e es ). •Low ope a ion expenses compa e o CT o MRI. •No ionizing adia ion like a CT. B. Accu acy o Measu emen As men ioned abo e, esul ing olume is compu ed om h ee-dimensional model o objec su ace. When using poin - cloud o m and Signed Volume o Te ahed on Me hod o ol- ume compu a ion, he only indispensable sou ce o unce ain y is hen unce ain y o measu ing he posi ion o single poin 4MRI and CT p inciples do no allow o use a poin -cloud o m o s o ing in o ma ions, he g id- ype memo y mus be used, wha leads o dec ease o me hod accu acy. 5Voxel is he smalles a ailable olume elemen in g id ep esen a ion. In mos ealiza ions, i is a cube wi h leng h o side gi en by spa ial esolu ion. 28 A. CHROMY Fig. 5. Volume ic unce ain y: De i ing heo e ical accu acy om lase scanning unce ain y. in 3D model (∆Xmax). As long as he Robo ic 3D Scanne is p ope ly calib a ed, he only conside able sou ces in luencing scanne ’s accu acy a e obo ic manipula o accu acy (∆M) and lase scanne accu acy (∆S). Rela ion be ween hese symbols has been de i ed in [8] as ollows6: ∆Xmax = 3 (∆M+ ∆S)(1) Compu e model o objec is composed om many single poin s close o each o he on i s su ace, as shown on Fig. 5. E e y poin is measu ed wi h unce ain y ∆Xmax, so he maximal absolu e olume ic unce ain y ∆Vis an spa ial a ea ma ked on Fig. 5 wi h dashed lines and exp essed as: ∆V=S·∆Xmax (2) whe e Sis su ace a ea o he measu ed objec [m2]. Rela i e accu acy (δV) is hen de ined as: δV=S V·∆Xmax (3) whe e Vis olume o he measu ed objec [m3]. Since Robo ic 3D Scanne is wo king wi h a ious manipula o s and lase scanne s, also inal accu acy δV will a y o di e en cons i u ions. Fo body-pa olu- me ic measu emen s, we use lase scanne Mic oEpsilon ScanCONTROL2750-100 wi h accu acy ∆S=±0.027 mm [64] and obo ic manipula o Epson C3 wi h accu acy o end- poin placemen ∆M=±0.013 mm [65]. Acco ding o (1), Robo ic 3D Scanne ’s accu acy in his case is ∆Xmax = ±0.12 mm and acco ding o (3), eal olume ic measu emen unce ain y (δV ) o p oposed de ice is: δV =S V·1.2·10−4(4) VI. COMPARISON WITH STANDARD METHODS Rela i e accu acies o all compa ed me hods a e summa- ized in Table I. Since ela i e accu acy o se e al me hods is no a cons an , i is no possible o clea ly compa e hem 6This equa ion encapsula es also he unce ain y o Robo ic 3D Scanne calib a ion p ocedu e and i is a e y pessimis ic es ima ion – see [8]. TABLE I METHODS COMPARISON: RELATIVE ACCURACIES OF STANDARD METHODS AND THE NEW METHOD. Me hod Rela i e Accu acy δV F us um Sign Model 8 % Disc Model 6 % Wa e Displacemen Me hod om 2 % (a 2700 cm3) up o 8 % (a 25 cm3) Op oelec onic Volume e s 2 % (1000 −3000 cm3) 3D Ul asonog aphy (5.0·10−3·S/V ) % Magne ic Resonance Imaging (MRI) (1.0·10−3·S/V ) % Compu ed Tomog aphy (CT) (2.0·10−4·S/V ) % Robo ic 3D Scanne (1.2·10−4·S/V ) % di ec ly. Because he alue o accu acy is a unc ion o measu ed objec shape (mo e p ecisely unc ion o su ace- o- olume a io), compa ison will be made on wo e e ence objec s: •p ecise cuboid – o be able o e i y compu ed accu acy expe imen ally (see sec ion VII), •human hand – o be able o compa e p oposed me hod wi h s anda d ones in hei s anda d wo king condi ions. Rela i e accu acy p og essions o pa icula me hods a e isualized on Fig. 6 ( o e e ence cuboid) and on Fig. 7 ( o human hand). These p og essions encapsula e he in luence o su ace-a ea- o- olume a ios (S/V ) on ela i e accu acy o me hod. I is clea om bo h igu es, ha accu acy o Robo ic 3D Scanne each he bes alue in e e y si ua ion. The second mos accu a e me hod is compu ing he olume om CT cap- u ed model, bu due o high doses o ionizing adia ion, i is used e y a ely. The hi d bes me hod – Wa e Displacemen Volume y – does no ha e such se ious limi a ions. F om his eason i is he mos used s anda d me hod, bu i s accu acy is app oxima ely 4 imes lowe han accu acy o Robo ic 3D Scanne Volume y. VII. EXPERIMENTAL RESULTS Theo e ical es ima ion o accu acy de i ed in sec ion V-B was expe imen ally e i ied on se o e e ence objec s wi h Fig. 6. Me hods compa ison: Rela i e measu emen accu acy δVand i s dependency on size o cuboid (i s olume). APPLICATION OF HIGH-RESOLUTION 3D SCANNING IN MEDICAL VOLUMETRY 29 Fig. 7. Me hods compa ison: Rela i e measu emen accu acy δVand i s dependency on size o human hand (i s olume). known dimensions and olume. The i s e e ence objec was a p ecise cuboid om blackened s eel manu ac u ed by p ecision-enginee ing company and consequen ly e i- ied in op ical measu ing chambe . De ining dimensions o cuboid a e 29.996 ±0.001 mm,39.990 ±0.001 mm and 49.993±0.001 mm, wi h pa allelisms o co esponding planes 0.005±0.001 mm,0.008±0.001 mm and 0.007±0.001 mm, and wi h maximal pe pendicula i y 0.007 ±0.001 mm (ϑ= 22◦C, φ = 46%). Acco ding o [66], ue olume o he e e ence objec V e is: V e = 59.969 ±0.005 cm3(5) Volume o his e e ence objec was 10 imes measu ed by Robo ic 3D Scanne Volume ic Me hod. A each measu e- men , same scanning ajec o y, bu di e en o ien a ion o e e ence objec inside scanning a ea was used. The mean alue and unce ain y ype A [67] we e de e mined om measu ed da a: Vmeas = 60.43 ±0.30 cm3(6) Assuming he equa ion (4), heo e ical ela i e accu acy in case o measu emen o objec wi h his size is δ heo y = ±1.88 %, so he measu ed alue Vmeas should be compa ible wi h in e al de ined by ue alue V e and ange δ heo y: V heo y = 59.97 ±1.13 cm3(7) This expe imen p o es alidi y o equa ion (4) o simila - sized objec s since Vmeas is compa ible wi h V heo y. To be able o e i y wide ange o olumes beyond he p e- cise me al e e ence objec , we used also he bigge e e ence cuboids. These objec s we e less-p ecise and made om wood (due o manu ac u ing expenses), bu i s dimensions we e known oge he wi h unce ain y o dimension measu emen s. Resul s o hese measu emen s we e e alua ed exac ly he same way. All esul s o e i ica ion a e summa ized in Table II and Fig. 8. All he ele an esul s a e compa ible, so he heo e - ically de i ed ela i e accu acy de ined by equa ion (4) was e i ied and seems o be alid. Figu e 8 shows heo e ical accu acy (”Decla ed accu acy”) and ue accu acy7wi h e o ba s p esen ing a iance o mea- su emen s (”T ue accu acy”). En i e se o measu ed alues s ayed below he decla ed accu acy, wha shows, ha his accu acy es ima ion is e y pessimis ic, and eal accu acy seems o be be e (”Appa en eal accu acy”). TABLE II EXPERIMENTAL VERIFICATION OF METHOD ACCURACY. Re . V e δV V heo y Vmeas Compa- Obj. [cm3] [%] [cm3] [cm3] ibili y #1 59.97 1.88 60.0±1.1 60.4±0.3YES #2 536.01 0.89 536.0±4.8 535.1±2.6YES #3 973.59 0.73 973.6±7.1 974.8±2.3YES #4 1734.71 0.63 1734.7±11.0 1737.7±3.0YES #5 2540.44 0.59 2540.4±15.0 2544.4±3.0YES VIII. CONCLUSION Despi e apid de elopmen o 3D scanning, i s boom im- pac s mo e o less only echnically-o ien ed domains. Bu he e a e many oppo uni ies also in o he a eas, like a medicine. This pape analyses needs o medicine in sense o 3D scan- ning and in e ences equi emen s on medical 3D scanning. Acco ding his speci ica ion, medical 3D scanne has been de eloped and i s con ibu ion has been demons a ed on one o i s applica ions – medical olume y. P oposed no el me hod o olume ic measu emen s eaches up o he bes accu acy compa ed o all s anda d me hods. Acco ding o pessimis ic heo e ical e alua ions o pe o mance, i s ela i e accu acy a ies om ±1 % o ±0.5 %, depending on complexi y o scanned limb su ace, speci ically on he su ace-a ea- o- olume a io. Bu du ing 7compu ed as ela i e di e ence be ween ue ( e e ence) alue and mea- su ed alue Fig. 8. Compa ing e i ica ion measu emen s and heo e ical me hod accu acy. 30 A. CHROMY e i ying expe imen s, ela i e e o was e en 4 imes lowe . The only a ea, whe e his me hod is no sui able is measu e- men o objec s smalle han 1 cm3, whe e measu emen e o eaches up o ±10 %. In addi ion o high accu acy, p oposed Robo ic 3D Scanne is mul i-pu pose de ice ha ing wide a e o usage also ou side o medical domain. I can be used in escue obo ics [68] o expe imen al biology [69]. In u u e wo k, i could be de eloped wi h mo e senso s wi h da a usion [70], wha would allow no only moni o ing he oedema, bu also ind he cen e o in lamma ion due o augmen ed eali y [71]. REFERENCES [1] “Google ends,” h ps://www.google.com/ ends. [Online]. A ailable: <h ps://www.google.com/ ends> [2] “A oundup o how odays 3D p in ing echnology is p og essing.” [Online]. A ailable: <h p://3dp in .com/ 27954/3d-p in ing- oundup/> [3] “Digi al unde go und,” h p://www.me museum.o g/abou - he-museum/museum-depa men s/o ice-o - he-di ec o /digi al- media-depa men /digi al-unde g ound. [Online]. A ailable: <h p://www.me museum.o g/abou - he-museum/ museum-depa men s/o ice-o - he-di ec o / digi al-media-depa men /digi al-unde g ound> [4] J. Pa k, “Digi al es o a ion o seokgu am g o o: The digi al a chi ing and he exhibi ion o sou h ko ea’s ep esen a i e UNESCO wo ld he - i age,” in P oceedings o 2012 In e na ional Symposium on Ubiqui ous Vi ual Reali y, ISUVR 2012, 2012, pp. 26–29. [5] L. Zalud, L. Kopecny, and F. Bu ian, “O pheus econnissance obo s,” in P oceedings o he 2008 IEEE In e na ional Wo kshop on Sa e y, Secu i y and Rescue Robo ics, SSRR 2008, Sendai, Japan, 2008, pp. 31–34. [6] L. Zalud, “ARGOS - sys em o he e ogeneous mobile obo eleope a- ion,” in 2006 IEEE/RSJ In e na ional Con e ence on In elligen Robo s and Sys ems, IROS 2006, Beijing; China, 2006, pp. 211–216. [7] J. Hu, Y. Cao, T. Wu, D. Li, and H. Lu, “High- esolu ion h ee- dimensional isualiza ion o he a spinal co d mic o ascula u e by synch o on adia ion mic o-CT,” Medical Physics, ol. 41, no. 10, 2014. [8] A. Ch omy and L. Zalud, “Robo ic 3D scanne as an al e na i e o s anda d modali ies o medical imaging,” Sp inge Plus, ol. 3, no. 1, p. 13, 2014. [Online]. A ailable: <h p://www.sp inge plus. com/con en /3/1/13> [9] S. Klein, M. A e y, G. Adams, S. Polla d, and S. Simske, “F om scan o p in : 3D p in ing as a means o eplica ion,” HP Labo a o ies Technical Repo , no. 30, 2014. [10] F. Haase, C. Siewe , D. B. on Rau en eld, J. U. Fischbach, and H. Sei e , “Compa ison o di e en me hods o quan i y he olume o ho se limbs,” Be line Und Mnchene Tie z liche Wochensch i , ol. 122, no. 3-4, pp. 126–131, Ap . 2009, PMID: 19350812. [11] S. H. Ridne , L. D. Mon gome y, J. T. Hepwo h, B. R. S ewa , and J. M. A me , “Compa ison o uppe limb olume measu emen echniques and a m symp oms be ween heal hy olun ee s and indi iduals wi h known lymphedema,” Lymphology, ol. 40, no. 1, pp. 35–46, Ma . 2007, PMID: 17539463. [12] R. C. B. Ribei o, S. M. P. F. Lima, A. C. G. Ca ei a, D. Masie o, and T. R. Chamlian, “In e - es e eliabili y assessmen o he olume ic measu emen o he hand in subjec s wi hou any changes in hei uppe ex emi ies,” Ac a Fisia ica, ol. 17, no. 1, pp. 3–7, 2010. [Online]. A ail- able: <h p://www.ac a isia ica.o g.b /audiencia_ pd .asp?aid2=68&nomeA qui o=en_ 17n1a02.pd > [13] D. M. Kaulesa Sukul, P. T. den Hoed, E. J. Johannes, R. an Dolde , and E. Benda, “Di ec and indi ec me hods o he quan i ica ion o leg olume: compa ison be ween wa e displacemen olume y, he disk model me hod and he us um sign model me hod, using he co ela ion coe icien and he limi s o ag eemen ,” Jou nal o Biomedical Enginee ing, ol. 15, no. 6, pp. 477–480, No . 1993, PMID: 8277752. [14] P. Konecny, “No e endy neu o ehabili aci [abs ac ],” in Dobsak P (ed) P oceedings o he IV. Dny Fyzio e apie: 11-12 Oc obe 2013, 2013, pp. 7 – 7. [15] K. La elle and D. B. S an on, “Measu emen o edema in he hand clinic,” May 2014. [16] J. M. A me and S. H. Ridne , “Measu emen echniques in assessmen o lymphedema,” Lymph Link A icle Rep in , ol. 18, no. 3, pp. 1–4, Sep. 2006. [17] F. B ijke , Y. F. Heijd a, F. J. Van Den Elshou , F. H. Bosch, and H. T. Folge ing, “Volume ic measu emen s o pe iphe al oedema in clinical condi ions,” Clinical Physiology (Ox o d, England), ol. 20, no. 1, pp. 56–61, Jan. 2000, PMID: 10651793. [18] T. Del ombe, J. Jama , S. Recloux, C. Leg and, N. Vandenb oeck, S. Theys, and P. Hanson, “Reliabili y and limi s o ag eemen o ci cum e en ial, wa e displacemen , and op oelec onic olume y in he measu emen o uppe limb lymphedema,” Lymphology, ol. 40, no. 1, pp. 26–34, Ma . 2007, PMID: 17539462. [19] P. Ka akas and M. G. Bozki , “An h opome ic indices in ela- ion o o e weigh and obesi y among u kish medical s uden s,” A chi es o Medical Science, ol. 8, no. 2, pp. 209–213, Ap . 2012, WOS:000304232700005. [20] R. Dams a, Diagnos ic and he apeu ical aspec s o lymphedema. D ach en; Maas ich : S ich ing Lym ologie Cen um Nede land (SLCN) ; Uni e si y Lib a y, Uni e si ei Maas ich [hos ], 2009. [21] A. R. Webb, In oduc ion o biomedical imaging. Hoboken, New Je sey: Wiley, 2003. [22] J. K. Udupa and G. T. He man, 3D Imaging in Medicine, Second Edi ion. CRC P ess, Sep. 1999. [23] J. W. Ramsay, P. J. Ba ance, T. S. Buchanan, and J. S. Higginson, “Pa e ic muscle a ophy and non-con ac ile issue con en in indi idual muscles o he pos -s oke lowe ex emi y,” Jou nal o Biomechanics, ol. 44, no. 16, pp. 2741–2746, No . 2011. [Online]. A ailable: <h p://linkinghub.else ie .com/ e ie e/pii/S0021929011005884> [24] M. d. A. Sil a-Cou o, C. L. P ado-Medei os, A. B. Oli ei a, C. C. Alcan a a, A. T. Guima aes, T. d. F. Sal ini, R. Ma ioli, and T. L. de Russo, “Muscle a ophy, olun a y ac i a ion dis u bances, and low se um concen a ions o IGF-1 and IGFBP-3 a e associa ed wi h weak- ness in people wi h ch onic s oke,” Physical The apy, ol. 94, no. 7, pp. 957–967, Jul. 2014, WOS:000338170100006. [25] W. Chong and P. Sidhu, Measu emen in Ul asound: A p ac ical handbook. CRC P ess, Ap . 2004. [26] P. Bening on, “Masse e muscle olume measu ed using ul asonog aphy and i s ela ionship wi h acial mo phology,” The Eu opean Jou nal o O hodon ics, ol. 21, no. 6, pp. 659–670, Dec. 1999. [Online]. A ailable: <h p://ejo.oupjou nals.o g/cgi/doi/10. 1093/ejo/21.6.659> [27] C. H. McCollough and F. E. Zink, “Pe o mance e alua ion o a mul i- slice CT sys em,” Medical Physics, ol. 26, no. 11, pp. 2223–2230, No . 1999, PMID: 10587202. [28] S a ni u ad p o jade nou bezpecnos , “Vyhlaska o adiacni och ane,” 2002. [29] Z. Seidl and M. Vanko , Magne ick ezonance hla y, mozku a p ee, a icenum ed. G ada, 2007. [Online]. A ailable: <h p: //www.ma inus.cz/?uI em=32926> [30] “Coo dina e measu ing machine his o y - i y yea s o CMM his o y leading up o a measu ing e olu ion.” [Online]. A ailable: <h p://www.coo d3-cmm. com/50-yea s-o -coo dina e-measu ing-machine/ indus y-de elopmen s-and-his o y/> [31] J. Shan and C. K. To h, Eds., Topog aphic lase anging and scanning: p inciples and p ocessing. Boca Ra on: CRC P ess/Taylo & F ancis G oup, 2009. [32] K. B. Smi h and Y. F. Zheng, “Accu acy analysis o poin lase iangula ion p obes using simula ion,” Jou nal o Manu ac u ing Science and Enginee ing, ol. 120, no. 4, pp. 736–745, No . 1998. [Online]. A ailable: <h p://dx.doi.o g/10.1115/1. 2830214> [33] R. G. Do sch, G. Husle , and J. M. He mann, “Lase iangula ion: undamen al unce ain y in dis ance measu emen ,” Applied Op ics, ol. 33, no. 7, p. 1306, Ma . 1994. [Online]. A ailable: <h p://www.op icsin obase.o g/abs ac . c m?URI=ao-33-7-1306> [34] R. Heinkelmann and H. Schuh, “Ve y long baseline in e e ome y: accu acy limi s and ela i is ic es s,” in Rela i i y in Fundamen al As onomy: Dynamics, Re e ence F ames, and Da a Analysis, se . P oceedings o he In e na ional As onomical Union, ol. 5, Ap . 2009, p. 286290. [Online]. A ailable: <h p://jou nals. camb idge.o g/a icle_S1743921309990524> APPLICATION OF HIGH-RESOLUTION 3D SCANNING IN MEDICAL VOLUMETRY 31 [35] R. Mo ano, C. Oz u k, R. Conn, S. Dubin, S. Zie z, and J. Nissano, “S uc u ed ligh using pseudo andom codes,” IEEE T ansac ions on Pa e n Analysis and Machine In elligence, ol. 20, no. 3, pp. 322–327, Ma . 1998. [36] K. Liu, Y. Wang, D. L. Lau, Q. Hao, and L. G. Hasseb ook, “Dual- equency pa e n scheme o high-speed 3-D shape measu emen ,” Op ics Exp ess, ol. 18, no. 5, p. 5229, Ma . 2010. [Online]. A ailable: <h p://www.op icsin obase.o g/abs ac . c m?URI=oe-18-5-5229> [37] S. Zhang and P. S. Huang, “High- esolu ion, eal- ime h ee- dimensional shape measu emen ,” Op ical Enginee ing, ol. 45, no. 12, pp. 123 601–123 601–8, 2006. [Online]. A ailable: <h p: //dx.doi.o g/10.1117/1.2402128> [38] J. Baqe sad, J. Ca , T. Lunds om, C. Niez ecki, P. A i abile, and M. Sla e y, “Dynamic cha ac e is ics o a wind u bine blade using 3D digi al image co ela ion,” in P oc. SPIE 8348, Heal h Moni o ing o S uc u al and Biological Sys ems 2012, ol. 8348, San Diego, Cali o nia, 2012, pp. 83 482I–83 482I–9. [Online]. A ailable: <h p://dx.doi.o g/10.1117/12.915377> [39] G. F. Ma shall, Ed., Handbook o op ical and lase scanning, se . Op ical enginee ing. New Yo k: Ma cel Dekke , 2004, no. 90. [40] F. Walkowski, R. Johns on, and N. P ice, “Tex u e mapping o he Fas SCAN hand-held lase scanne ,” in Image and Vision Compu ing New Zealand, 2008. IVCNZ 2008. 23 d In e na ional Con e ence, No . 2008, pp. 1–6. [41] K. S obl, E. Mai , T. Bodenmulle , S. Kielho e , W. Sepp, M. Suppa, D. Bu schka, and G. Hi zinge , “The sel - e e enced DLR 3D-modele ,” in IEEE/RSJ In e na ional Con e ence on In elligen Robo s and Sys- ems, 2009. IROS 2009, Oc . 2009, pp. 21–28. [42] A. L. Reyes, J. M. Ce an es, and N. C. Gu i ez, “Low cos 3D scanne by means o a 1D op ical dis ance senso ,” P ocedia Technology, ol. 7, pp. 223–230, 2013. [Online]. A ailable: <h p://www.sciencedi ec .com/science/ a icle/pii/S2212017313000297> [43] Zhang, A., Hu, S., Chen, Y., Liu, H., Yang, F., and Liu, J., “Fas con inuous 360 deg ee colo 3D lase scanne ,” in INTERNATIONAL ARCHIVES OF PHOTOGRAMMETRY REMOTE SENSINGAND SPA- TIAL INFORMATION SCIENCES, ol. 1, 2008, pp. 409–414. [44] A. Ch omy and L. Zalud, “No el 3D modelling sys em cap u ing objec s wi h Sub-Millime e esolu ion,” Ad ances in Elec ical and Elec onic Enginee ing, ol. 12, no. 5, pp. 476–487, Dec. 2014. [Online]. A ailable: <h p://ad ances.u c.sk/index.php/ AEEE/a icle/ iew/1123> [45] R. M. Mu ay, Z. Li, S. S. Sas y, and S. S. Sas y, A Ma hema ical In oduc ion o Robo ic Manipula ion. CRC P ess, Ma . 1994. [46] S. Y. No , Handbook o Indus ial Robo ics. John Wiley & Sons, 1999. [47] F. Solc and L. Zalud, Robo ika. B no: B no Uni e si y o Technology, 2006. [48] L. G. B own, “A su ey o image egis a ion echniques,” ACM Compu . Su ., ol. 24, no. 4, p. 325376, Dec. 1992. [Online]. A ailable: <h p://doi.acm.o g/10.1145/146370.146374> [49] P. Besl and N. D. McKay, “A me hod o egis a ion o 3-D shapes,” IEEE T ansac ions on Pa e n Analysis and Machine In elligence, ol. 14, no. 2, pp. 239–256, Feb. 1992. [50] S. Rusinkiewicz and M. Le oy, “E icien a ian s o he ICP algo i hm,” in Thi d In e na ional Con e ence on 3-D Digi al Imaging and Modeling, 2001. P oceedings, 2001, pp. 145–152. [51] Alshawa, M., “ICL: i e a i e closes line - a no el poin cloud egis a- ion algo i hm based on linea ea u es,” Ekscen a , no. 10, pp. 53–59, 2007. [52] J. Poppinga, N. Vaske icius, A. Bi k, and K. Pa hak, “Fas plane de ec ion and polygonaliza ion in noisy 3D ange images,” in IEEE/RSJ In e na ional Con e ence on In elligen Robo s and Sys ems, 2008. IROS 2008, Sep. 2008, pp. 3378–3383. [53] F. Bellocchio, N. A. Bo ghese, S. Fe a i, and V. Piu i, 3D Su ace Recons uc ion. New Yo k, NY: Sp inge New Yo k, 2013. [Online]. A ailable: <h p://link.sp inge .com/10.1007/ 978-1-4614-5632-2> [54] K. S obl, E. Mai , and G. Hi zinge , “Image-based pose es ima ion o 3-D modeling in apid, hand-held mo ion,” in 2011 IEEE In e na ional Con e ence on Robo ics and Au oma ion (ICRA), May 2011, pp. 2593– 2600. [55] B. Meh a and R. Ma inescu, “Compa ison o image gene a ion and p ocessing echniques o 3D econs uc ion o he human skull,” in P oceedings o he 23 d Annual In e na ional Con e ence o he IEEE Enginee ing in Medicine and Biology Socie y, 2001, ol. 4, 2001, pp. 3687–3690 ol.4. [56] D. Chwa, J. Kang, and J. Choi, “Online ajec o y planning o obo a ms o in e cep ion o as maneu e ing objec unde o que and eloci y cons ain s,” IEEE T ansac ions on Sys ems, Man and Cybe ne ics, Pa A: Sys ems and Humans, ol. 35, no. 6, pp. 831–843, No . 2005. [57] J. Le inson, J. Askeland, J. Becke , J. Dolson, D. Held, S. Kammel, J. Kol e , D. Lange , O. Pink, V. P a , M. Sokolsky, G. S anek, D. S a ens, A. Teichman, M. We ling, and S. Th un, “Towa ds ully au onomous d i ing: Sys ems and algo i hms,” in 2011 IEEE In elligen Vehicles Symposium (IV), Jun. 2011, pp. 163–168. [58] G. Xiaoqing and W. Jidong, “T ajec o y planning heo y and me hod o indus ial obo ,” in 2011 3 d In e na ional Con e ence on Compu e Resea ch and De elopmen (ICCRD), ol. 2, Ma . 2011, pp. 340–343. [59] R. Danescu, “Obs acle de ec ion using dynamic Pa icle-Based oc- cupancy g ids,” in 2011 In e na ional Con e ence on Digi al Image Compu ing Techniques and Applica ions (DICTA), Dec. 2011, pp. 585– 590. [60] N. Fai ield and D. We e g een, “E idence g id-based me hods o 3D map ma ching,” in IEEE In e na ional Con e ence on Robo ics and Au oma ion, 2009. ICRA ’09, May 2009, pp. 1637–1642. [61] H. Yang and J. Chen, “Poin cloud da a enhancemen based on laye con- nec ed egion,” in 2014 In e na ional Con e ence on Audio, Language and Image P ocessing (ICALIP), Jul. 2014, pp. 600–604. [62] A. Ch omy, P. Kocmano a, and L. Zalud, “C ea ing Th ee-Dimensional compu e models using obo ic manipula o and lase scanne s,” in 12 h IFAC Con e ence on P og ammable De ices and Embedded Sys ems (2013), se . P og ammable de ices and sys ems. Velke Ka lo ice: Else ie B.V., Sep. 2013, pp. 268–273. [Online]. A ailable: <h p: //www.i ac-pape sonline.ne /De ailed/62529.h ml> [63] C. Zhang and T. Chen, “E icien ea u e ex ac ion o 2D/3D objec s in mesh ep esen a ion,” in 2001 In e na ional Con e ence on Image P ocessing, 2001. P oceedings, ol. 3, 2001, pp. 935–938 ol.3. [64] Mic o-Epsilon, “Ins uc ion manual scanCONTROL,” 2008. [Online]. A ailable: <h p://www.mic o-epsilon.cz/download/ manuals/man--scanCONTROL-2700--en.pd > [65] Epson Robo s, “Epson c3 compac 6-Axis Robo Manual,” 2011. [Online]. A ailable: <h p:// obo s.epson.com/admin/ uploads/p oduc _ca alog/ iles/EPSON_C3_Robo _ Manual(R7).pd > [66] T. Jame son, “Unce ain y example using simple p opaga ion o unce - ain y ules,” Aug. 2009. [67] R. Palenca , F. Vdolecek, and M. Halaj, “Nejis o y me eni ii: nejis o y p imych me eni,” Au oma, ol. 2001, no. 10, pp. 55–56, 2001. [68] L. Zalud, L. Kopecny, F. Bu ian, and T. Flo ian, “CASSANDRA - he - e ogeneous econnaissance obo ic sys em o dange ous en i onmen s,” in 2011 IEEE/SICE In e na ional Symposium on Sys em In eg a ion, SII 2011, 2011, pp. 1275–1280. [69] L. Nejdl, J. Kud , K. Cihalo a, D. Chudobo a, M. Zu ek, L. Zalud, L. Kopecny, F. Bu ian, B. Ru kay-Nedecky, S. K izko a, M. Konecna, D. Hynek, P. Kopel, J. P asek, V. Adam, and R. Kizek, “Remo e- con olled obo ic pla o m ORPHEUS as a new ool o de ec ion o bac e ia in he en i onmen ,” Elec opho esis, ol. 35, no. 16, pp. 2333– 2345, 2014. [70] L. Zalud and P. Kocmano a, “Fusion o he mal imaging and CCD came a-based da a o s e eo ision isual elep esence,” in 2013 IEEE In e na ional Symposium on Sa e y, Secu i y, and Rescue Robo ics, SSRR 2013, 2013. [71] P. Kocmano a, L. Zalud, and A. Ch omy, “3D p oximi y lase scanne calib a ion,” in 2013 18 h In e na ional Con e ence on Me hods and Models in Au oma ion and Robo ics (MMAR), Aug. 2013, pp. 742–747.