CRANIOMAXILLOFACIAL TRAUMA

1 57
2 58
3 59
4
5
Template-Based Orbital Wall Fracture 60
61
6
7 Treatment Using Statistical 62
63
8 64
9 Shape Analysis 65
10 66
11 Q5 Hans-Martin Doerfler, MEng,* Heike Huempfner-Hierl, MD, DDS, PhD,y 67
12 Daniel Kruber, Dipl-Inf,z Peter Schulze, PhD-Eng,x and Thomas Hierl, MD, DDS, PhDk 68
13 69
Purpose: Aim of this study was to investigate whether a mold generated from a statistical shape model of
14 70
the orbit could be generated to provide a cost-efficient means for the treatment of orbital fractures.
15 71
16 Materials and Methods: A statistical shape model was created from 131 computed tomographic (CT) 72
17 scans of unaffected adult middle European human orbits. To generate the model, CT scans were 73
18 segmented in Brainlab software, preregistered using anatomic landmarks, trimmed to an identical size, 74
19 and definitely registered. Then, the model was created using the global master algorithm. Based on this 75
20 model, a mold consisting of a male part and a female part was constructed and printed using a rapid pro- 76
21 totyping technique. 77
22 Results: A statistical shape model of the human orbit was generated from 125 CT scans. Six scans (4.5%) 78
23 presented major anatomic deviations and were discarded. A solid mold based on this model was printed. 79
24 Using this mold, flat titanium mesh could be successfully deformed to serve as an orbital implant. 80
25 81
Conclusion: A mold based on the statistical orbital shape could serve as a cost-effective means for the
26 82
treatment of orbital fractures. It allows the anatomic preformation of titanium or resorbable implant ma-
27 83
terial for orbital reconstruction. Because these materials could be cut from larger sheets, the use of a mold
28 84
Q3 would be a cost-effective treatment alternative.
29 85
Ó 2017 Published by Elsevier Inc on behalf of the American Association of Oral and Maxillofacial
30 86
Surgeons
31 87
J Oral Maxillofac Surg -:1.e1-1.e8, 2017
32 88
33 89
34 The human orbit is anatomically complex and de- Materials used range from autologous bone grafts,3 90
35 mands precise reconstruction of fractures to preserve intraoperatively bent titanium meshes,4 preformed 91
36 esthetics and function. In recent decades, a multitude titanium meshes,5,6 to patient-specific implants 92
37 of materials and procedures has been suggested.1,2 (PSIs) generated from bioceramics, polyethylene, or ti- 93
38 Concomitant with the technical progress in image tanium.7-10 As an alternative to PSIs, alloplastic 94
39 acquisition and implant manufacturing has been a materials have been preoperatively modified using 95
40 trend toward the optimal fit of orbital implants. patient models created by rapid prototyping.11,12 96
41 97
42 98
43 *Engineer, Department of Oral and Maxillofacial Plastic Surgery, the German Federal Ministry of Economics and Technology (ZIM 99
44 Leipzig University and University of Applied Sciences (HTWK), KF2036708SS0). 100
45 Leipzig, Germany. Conflict of Interest Disclosures: None of the authors have a rele- 101
Q2
46 ySenior Consultant, Department of Oral and Maxillofacial Plastic vant financial relationship(s) with a commercial interest. 102
47 Surgery, Leipzig University, Leipzig, Germany. Address correspondence and reprint requests to Dr Hierl: 103
48 zComputer Scientist, Department of Oral and Maxillofacial Plastic Department of Oral and Maxillofacial Plastic Surgery, Leipzig 104
49 Surgery, Leipzig University, Leipzig, Germany. University, Liebigstrasse 12, Leipzig 04103, Germany; e-mail: 105
50 xFaculty of Mechanical and Energy Engineering, University of Thomas.Hierl@medizin.uni-leipzig.de 106
51 Applied Sciences (HTWK), Leipzig, Germany. Received January 27 2017 107
52 kSenior Consultant, Department of Oral and Maxillofacial Plastic Accepted March 27 2017 108
53 Surgery, Leipzig University, Leipzig, Germany. Ó 2017 Published by Elsevier Inc on behalf of the American Association of Oral 109
54 This investigation was funded by the German Federal Ministry of and Maxillofacial Surgeons 110
55 Economics and Technology (ZIM KF 2036713AK2). The develop- 0278-2391/17/30366-X 111
56 ment of the 3-dimensional analysis software was funded I part by http://dx.doi.org/10.1016/j.joms.2017.03.048 112

1.e1
FLA 5.4.0 DTD
YJOMS57738_proof
26 April 2017
2:44 pm
CE AH
 1.e2 TEMPLATE FOR ORBITAL WALL FRACTURE TREATMENT Q1

113 Furthermore, intraoperative navigation has been used were manually modified to the anatomically correct 169
114 to increase surgical precision irrespective of the shape within Brainlab software. Manual correction 170
115 chosen implant material.13,14 The technical progress included using the ‘‘paint’’ and ‘‘erase’’ tools for 171
116 from flat and individually contoured titanium meshes cleanup and the ‘‘wand tool’’ for major deviations. 172
117 to solid and intraoperatively barely adjustable PSIs Next, all orbits were exported in stereolithographic 173
118 also has caused a massive increase in costs for orbital (STL) file format. Nine scans were omitted because 174
119 implants. Pre-contoured titanium meshes are an inter- of quality problems in the exported STL file; thus, 175
120 mediate state between flat sheet meshes and PSIs. In a 131 orbits remained. These had been obtained from 176
121 previous study, the authors investigated the different 38 women and 93 men 18 to 68 years old. Three 177
122 shapes of commercially available pre-bent titanium anatomic landmarks were set to serve for the 178
123 meshes15 and noticed major differences in their shape preregistration (anterior inferior orbital fissure, 179
124 and conformance with a self-created statistical orbital inferior optic foramen, and inferior lacrimal groove) 180
125 model. Therefore, this study investigated whether a and the orbits were coarsely trimmed as presented 181
126 cost-saving but nevertheless anatomically correct in Figure 1. 182
127 method in the treatment of orbital floor fractures could The resulting orbital shapes were preregistered in 183
128 be established. Such a method should be superior to 3matic software (Materialise, Gilching, Germany). The 184
129 conventional sheet titanium meshes and more flexible next step was a fine trimming using 3matic software to 185
130 than existing pre-contoured meshes. In this study, a create an almost identical size of all 131 orbital areas. 186
131 mold and plunger based on a model generated by sta- Then, all orbits were registered in 3matic using N-points 187
132 tistical shape analysis was judged preferable. Thus, registration and global registration. Then, the statistical 188
133 flexibility of the used material and the needed implant model was created. Three ways to do so were evaluated. 189
134 size would be combined with the closest anatomic fit The first way was Procrustes registration, the generation 190
135 compared with the gold standard of PSIs. of semi-landmarks, decrease of semi-landmark energy by 191
136 sliding movements, computation of new landmarks by 192
137 Materials and Methods principal component analysis, and creating the final 193
138 model.18,19 This would have necessitated the use of 194
139 Two steps were needed to create the mold and several software packages and different operating 195
140 plunger: 1) the generation of a model based on statis- system platforms; thus, the entire algorithm was 196
141 tical shape analysis and 2) the construction and fabri- programmed using VTK (Visualization ToolKit, 197
142 cation of the device. Kitware, Clifton Park, NY), which served as the second 198
143 software strategy. The third way was using the best-fit 199
CREATING A STATISTICAL SHAPE MODEL
144 and global master algorithms implemented in GOM 200
145 A model of the full orbit with plunger-type male and Inspect 8 (Atos Professional 7.5SR2, GOM, Braunsch- 201
146 female molds was impractical because it would have weig, Germany).20,21 Because the third method turned 202
147 shown undercuts. Therefore, a model of the typical out to be the fastest and easiest, it was chosen. 203
148 fracture sites, namely the inferior and medial walls, 204
149 was chosen. The generation consisted of the following 205
150 steps: GENERATION OF THE PLUNGER AND EVALUATION 206
151 OF THE RESULTING SHAPE 207
152 - Segmentation of bony orbits After the statistical shape model was generated, a 208
153 - Removing surplus areas and placing anatomic male mold and a female mold had to be constructed 209
154 landmarks for preregistration using Catia 5 R2012 SP4 (Catia 5, Dassault Syst
eme, 210
155 - Preregistration using landmarks and definite regis- Paris, France). To test the effectiveness of the mold, 211
156 tration of surfaces 2 flat titanium meshes were chosen randomly and 212
157 - Generation of the statistical shape model deformed and the resulting shape was compared 213
158 - Clinical analysis and final modification if necessary with the statistical shape model. The authors used 214
159 the same approach as in their previous comparison 215
160 One hundred forty computed tomographic (CT) of the statistical shape model with commercially avail- 216
161 scans of adult patients of local origin (ie, middle Euro- able pre-bent meshes.15 The meshes (Stryker, Frei- 217
162 pean ethnicity) with unaffected orbits were chosen for burg, Germany; and DePuy Synthes, Zuchwil, 218
163 the model. All were segmented using the atlas-based Switzerland) were cut to the size of the template and 219
164 automatic segmentation algorithm of iPlan software deformed. Then, the meshes were coated with ultra- 220
165 (Brainlab, Munich, Germany). Because previous thin flexible tape, sprayed with matt white paint, 221
166 studies have shown that the left and right sides do and digitized using a high-resolution industrial scanner 222
167 not show statistically relevant differences,16,17 only 1 (0.015-mm resolution; ATOS IIe revised 01, GOM). 223
168 side was selected. If necessary, the generated files The digitized meshes were compared with the 224

FLA 5.4.0 DTD
YJOMS57738_proof
26 April 2017
2:44 pm
CE AH

 DOERFLER ET AL 1.e3

225 281
226 282
227 283
228 284
229 285
230 286
231 287
232 288
233 289
234 290
235 291
236 292
237 293
238 294
web 4C=FPO

239 295
240 296
241 297
242 298
243 FIGURE 1. Process used to generate orbital areas for the statistical model. Left, The stereolithographic file of the orbit exported from Brainlab 299
iPlan software. Blue areas delineate removed areas (outside removed). Yellow areas depict typical areas of orbital wall fractures. Green areas
244 denote the transition from the medial wall to the roof. Red areas denote points used for preregistration. Right, Trimmed surface.
300
245 301
Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral Maxillofac Surg 2017.
246 302
247 statistical shape model after registration using the iter- walls was generated. Comparison of all 131 orbits with 303
248 ative closest-point algorithm and differences were dis- the resulting model for tolerance and deviation showed 304
249 played in a color-coded distance map. that 6 orbits (4.5%) were outliers in exhibiting major dif- 305
250 This study was approved by the local ethics commit- ferences. Figure 2 shows a case of typical outlier geom- 306
251 tee (Medical Faculty, Leipzig University, Leipzig, Ger- etry, and Figure 3 presents a good anatomic fit. 307
252 many). According to the ethics committee decision, When the outliers were analyzed, no correlation 308
253 no consent for participation was needed. This study with age, gender, or specific anatomic regions could 309
254 was performed according to the Declaration of Helsinki. be found. Therefore, a new model without these out- 310
255 liers was generated, consisting of 125 CT scans. 311
256 Figure 4 displays the final statistical shape model. 312
257 Results 313
Analysis of the final model showed that the greatest
258 Using the best-fit and global master algorithms, a sta- variation in the test population existed in the periph- 314
259 tistical shape model of the medial and inferior orbital ery of the model. Thus, the transition from the medial 315
260 316
261 317
262 318
263 319
264 320
265 321
266 322
267 323
268 324
269 325
270 326
271 327
272 328
273 329
274 330
web 4C=FPO

275 331
276 332
277 333
278 334
279 FIGURE 2. One example of the 6 outliers. Distinct shape differences in the lateral and posterior medial orbital floor and medial wall are visible. 335
280 Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral Maxillofac Surg 2017. 336

FLA 5.4.0 DTD
YJOMS57738_proof
26 April 2017
2:44 pm
CE AH

 1.e4 TEMPLATE FOR ORBITAL WALL FRACTURE TREATMENT

337 393
338 394
339 395
340 396
341 397
342 398
343 399
344 400
345 401
346 402
347 403
348 404
349 405
web 4C=FPO

350 406
351 407
352 408
353 409
354 FIGURE 3. Example of a good fit, with anatomy consistent with the statistical shape model. 410
355 Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral Maxillofac Surg 2017. 411
356 412
357 to the superior wall (especially near the optic fora- the plunger. Right and left molds were generated in 413
358 men), the transition from the inferior to the lateral solid ABS plastic with a rapid prototyping technique 414
359 wall, and the inferior orbital rim showed the greatest (FDM Titan, Stratasys Co, Rheinm€ unster, Germany) us- 415
360 variation (Fig 5). However, the typical fracture sites ing a 0.2-mm nozzle for fine printing (Figs 6, 7). 416
361 presented good interpatient conformity. As an alternative to the flat titanium meshes, resorb- 417
362 After the statistical shape model was generated, a able or polyethylene-coated meshes could be inserted. 418
363 male mold and a female mold were constructed in Furthermore, orbital size meshes could be cut from 419
364 Catia 5. The surface was arranged to prohibit under- larger rectangular mesh sheets, which are available 420
365 cuts, and guide bars were created on the left and right in various strengths and mesh geometries from several 421
366 sides to guarantee a straight downward movement of companies. In the mold references could be integrated 422
367 to help cut the desired size of the implant. The mold 423
368 could be sterilized and thus the device could be used 424
369 intraoperatively. Because it would have been impos- 425
370 sible to evaluate all possible materials and sizes for con- 426
371 formity of the resulting shape with the mold, 2 meshes 427
372 were chosen randomly for evaluation. The evaluation 428
373 was performed by choosing conformity corridors, 429
374 that is, stating the percentage of the area within a given 430
375 deviation. In the corridor of 0.2-mm deviation for the 431
376 statistical shape model, 81% of the Stryker mesh was 432
377 within these boundaries. Maximum deviation was 433
378 0.34 mm in the lateral anterior border (Fig 8). For 434
379 the Synthes mesh, 71% of its area was within the 435
380 0.2-mm deviation corridor. The maximum discrep- 436
381 ancy was found in the anterior lateral region (0.6 mm). 437
382 438
web 4C=FPO

383 439
Discussion
384 440
385 For the creation of the statistical shape model, the 441
386 specimen was compared with pre-existing models. 442
387 FIGURE 4. The final statistical shape model generated from 125 To the authors’ knowledge,15 3 companies offer 7 443
388 left orbital computed tomographic scans. FOI, fissura orbitalis preformed orbital titanium meshes. Two companies 444
inferior; IR, infraorbital rim; OF, optical foramen; TIL, transition
389 from inferior to lateral wall; TIM, transition from inferior to medial
offer large and small meshes that differ only in size 445
390 wall; TMS, transition from medial to superior wall. and not in shape, whereas 1 company offers large 446
391 Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral and small implants with different shapes. For size of 447
392 Maxillofac Surg 2017. the patient group and the procedures chosen by 448

FLA 5.4.0 DTD
YJOMS57738_proof
26 April 2017
2:44 pm
CE AH

 DOERFLER ET AL 1.e5

449 505
450 506
451 507
452 508
453 509
454 510
455 511
456 512
457 513
458 514
459 515
460 516
461 517
462 518
463 519
464 520
465 521
web 4C=FPO

466 522
467 523
468 524
469 525
470 FIGURE 5. Display of range and standard deviation of the statistical shape orbital wall model. The greatest deviations occurred in the transition 526
471 from the medial to the superior wall, especially near the optic foramen. The second zone is the upturning of the caudal wall toward the lateral 527
wall. The infraorbital rim geometry is the third area with the greatest shape differences.
472 528
Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral Maxillofac Surg 2017.
473 529
474 530
475 these companies, KLS Martin (Tuttlingen, Germany) differences between Europeans and non-Euro- 531
476 and Zimmer-Biomet (Freiburg, Germany) have pro- peans,22,23 further studies would be necessary if a 532
477 vided no information. Stryker stated that their model similar approach were to be used in an non- 533
478 is based on 300 CT scans but provided no information European environment. For methods to generate a 534
479 on the patient group or on the way the model was statistical shape model, the most widely used proced- 535
480 generated. Only DePuy Synthes has provided more ure would be the first one discussed in the Materials 536
481 substantial information on their meshes.16,17 Sixty- and Methods. Digitizing a given specimen using 537
482 eight of 70 CT sets were included. Thus, 125 of 131 semi-landmarks has been used in numerous investiga- 538
483 CTs of patients for the present final model seem tions.18,19 To the authors’ knowledge, no studies have 539
484 acceptable for a given ethnicity. Because of anatomic 540
485 541
486 542
487 543
488 544
489 545
490 546
491 547
492 548
493 549
494 550
495 551
496 552
web 4C=FPO

497 553
web 4C=FPO

498 554
499 555
500 556
FIGURE 7. The plunger is inserted and the desired mesh is held
501 FIGURE 6. The resulting parts of the mold (left) with the lateral between the 2 parts. The mesh is deformed by manual downward
557
502 guiding wings for exact downward movement. pressure. 558
503 Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral 559
504 Maxillofac Surg 2017. Maxillofac Surg 2017. 560

FLA 5.4.0 DTD
YJOMS57738_proof
26 April 2017
2:44 pm
CE AH

 1.e6 TEMPLATE FOR ORBITAL WALL FRACTURE TREATMENT

561 prototyping mold. The present approach differs 617

562 because a generic mold was created that could serve 618
563 most patients presenting with inferior and medial 619
564 orbital wall fractures. The main advantage of a male 620
565 mold and a female mold (piston and plunger in the 621
566 present template) is to create a maximum adaption 622
567 of a given material to the needed specific shape. In 623
568 contrast to a mold, an anatomic statistical shape 624
569 model of the full orbit would cover all walls, but the 625
570 exact control of fit and manual adaptation would be 626
571 more difficult. This applies to the commercially 627
572 available orbital models, which are provided by KLS 628
web 4C=FPO

573 Martin and Medartis AG (Basel, Switzerland). In these 629

574 models adaptation will work by manual pressure or 630
575 using blunt rounded instruments or pliers. In 631
576 addition to the piston-and-plunger mechanism of the 632
577 FIGURE 8. Shape evaluation of a deformed flat titanium sheet present template, manual adaptation of implant mate- 633
(Stryker). Eight-one percent of the area is within a 0.2-mm
578 deviation corridor. Regions of maximum deviation are stated, rials in the present template is facilitated because a 634
579 reaching 0.34 mm in the anterior lateral area. male mold and a female mold are provided. The only 635
580 Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral known approach for a template of the inferior and 636
581 Maxillofac Surg 2017. medial walls serving all shape variations has been the 637
582 robotic press suggested by Metzger et al,27 which 638
583 has not been introduced into clinical practice. Con- 639
584 reported potential differences among methods to cerning possible implant materials, only spring hard 640
585 create statistical shape models, but because the meshes or materials with increased thickness would 641
586 global master algorithm has been used successfully be difficult to deform. Although most companies do 642
587 in previous medical studies,20,21 the present not state the rigidity of their meshes (tension free to 643
588 approach seems acceptable. Concerning the spring hard), spring hardness is the least likely to be 644
589 variation within the statistical shape model, the encountered. Because the present prototype mold is 645
590 present findings are similar to those of Kamer designed to work as a hand-operated device, titanium 646
591 et al16 and Noser et al.17 They reported on identical meshes thicker than 0.4 to 0.6 mm would be difficult 647
592 regions of greater variation, namely the transition to deform. The 2 implant materials tested showed 648
593 from the medial to the cranial wall, the transition that the resulting meshes were within clinically 649
594 from the inferior to the lateral wall, and the region acceptable differences for the mold. 650
595 of the optic foramen as major deviation regions and Because a small percentage of patients will present a 651
596 the infraorbital rim and posterior floor as minor devi- major anatomic deviation, a preoperative virtual try-in 652
597 ation areas. Nevertheless, the authors’ model shows seems reasonable. The STL file of the present statistical 653
598 differences as they previously reported in detail.15 orbital wall model might be used in surgical software 654
599 Different from Kamer et al16 and Noser et al,17 the au- such the Brainlab iPlan to examine the fit of the tem- 655
600 thors modified their initial model by removing plate. A typical example is shown in Figure 9. 656
601 distinct outliers. Almost 5% of the present patients ex- Thus, implant alternatives for major shape differ- 657
602 hibited scattered major differences in clinically crit- ences could be considered. 658
603 ical orbital regions. Thus, the authors created a This investigation showed that a statistical 659
604 model from all 131 patients rather than eliminate a anatomic model of the orbit is feasible because hu- 660
605 small group of nonuniform outliers to obtain an man orbits present a distinct conformity. A generic 661
606 even better fit in the vast majority. The authors mold to adapt different materials could be a cost- 662
607 believe that anatomically preformed implants will saving approach for most patients with orbital frac- 663
608 never cover all possible variations but are a rational ture compared with preformed implants or PSIs. 664
609 means for most patients. Thus, they abandoned the The present virtual model and the generated implant 665
610 6 outliers. They could not determine any pre- material can be used in combination with navigation 666
611 existing trauma or other causes, but unnoticed systems to show their concordance with the specific 667
612 trauma or inflammation cannot be ruled out. patient situation and to display possible modifica- 668
613 The idea to create a mold to deform orbital implants tions. Because the local ethics committee consented 669
614 has been applied to PSIs in some cases.24-26 In those only to the generation of the mold prototype but 670
615 cases, a mold was generated to deform a resorbable explicitly not to its clinical use, the authors have per- 671
616 or titanium mesh by an individually generated rapid formed only virtual try-ins in patients. However, these 672

FLA 5.4.0 DTD
YJOMS57738_proof
26 April 2017
2:44 pm
CE AH

 DOERFLER ET AL 1.e7

673 729
674 730
675 731
676 732
677 733
678 734
679 735
680 736
681 737
682 738
683 739
684 740
685 741
686 742
687 743
688 744
689 745
690 746
691 747
692 748
693 749
694 750
695 751
696 752
web 4C=FPO

697 753
698 754
699 755
700 756
701 FIGURE 9. Presurgical virtual try-in of the template created in Brainlab iPlan software (2-wall fracture case). This patient was not included in the 757
scans used for the model. The template shows an exact anatomic fit.
702 758
Doerfler et al. Template for Orbital Wall Fracture Treatment. J Oral Maxillofac Surg 2017.
703 759
704 760
705 761
suggest that the present approach would be effective ethylene orbital wall implant. J Craniomaxillofac Surg 41:
706 282, 2013 762
Q4 in clinical practice. 9. Stoor P, Suomalainen A, Lindqvist C, et al: Rapid prototyped
707 763
patient specific implants for reconstruction of orbital wall
708 defects. J Craniomaxillofac Surg 42:1644, 2014
764
709 References 10. Gander T, Essig H, Metzler P, et al: Patient specific implants 765
710 (PSI) in reconstruction of orbital floor and wall fractures. 766
1. Baino F: Biomaterials and implants for orbital floor repair. Acta J Craniomaxillofac Surg 43:126, 2015
711 11. Kozakiewicz M, Elgalal M, Loba P, et al: Clinical application of 3D
767
Biomater 9:3248, 2011
712 2. Dubois L, Steenen SA, Gooris PJJ, et al: Controversies in orbital pre-bent titanium implants for orbital floor fractures. J Cranio- 768
713 reconstruction—I. Defect-driven orbital reconstruction: A sys- maxillofac Surg 37:229, 2009 769
tematic review. Int J Oral Maxillofac Surg 44:308, 2015 12. Mustafa SF, Evans PL, Bocca A, et al: Customized titanium recon-
714 struction of post-traumatic orbital wall defects: A review of 22
770
3. Kronig SAJ, van der Mooren RJG, Strabbing EM, et al: Pure orbital
715 blowout fractures reconstructed with autogenous bone grafts: cases. Int J Oral Maxillofac Surg 40:1357, 2011 771
716 Functional and aesthetic outcomes. Int J Oral Maxillofac Surg 13. Schmelzeisen R, Gellrich NC, Schoen R, et al: Navigation-aided 772
45:507, 2016 reconstruction of medial orbital wall and floor contour in
717 cranio-maxillofacial reconstruction. Injury 35:955, 2004
773
4. Gabrielli MF, Monnazzi MS, Passeri LA, et al: Orbital wall recon-
718 struction with titanium mesh: Retrospective study of 24 pa- 14. Dubois L, Schreurs R, Jansen J, et al: Predictability in orbital recon- 774
719 tients. Craniomaxillofac Trauma Reconstr 4:151, 2011 struction: A human cadaver study. Part II: Navigation-assisted 775
5. Gordon CR, Susarla SM, Yaremchuk MJ: Quantitative assessment orbital reconstruction. J Craniomaxillofac Surg 43:2042, 2015
720 15. Huempfner-Hierl H, Doerfler HM, Kruber D, et al: Morphologic
776
of medial orbit fracture repair using computer-designed anatom-
721 ical plates. Plast Reconstr Surg 130:698e, 2012 comparison of preformed orbital meshes. J Oral Maxillofac Surg 777
722 6. Scolozzi P, Momjian A, Heuberger J, et al: Accuracy and predict- 73:1119, 2015 778
723 ability in use of AO three-dimensionally preformed titanium 16. Kamer L, Noser H, Schramm A, et al: Orbital form analysis: 779
mesh plates for posttraumatic orbital reconstruction: A pilot Problems with design and positioning of precontoured
724 study. J Craniofac Surg 20:1108, 2009 orbital implants. A serial study using post-processed clinical 780
725 7. Klein M, Glatzer C: Individual CAD/CAM fabricated glass- CT data in unaffected orbits. Int J Oral Maxillofac Surg 39: 781
726 bioceramic implants in reconstructive surgery of the bony 666, 2010 782
orbital floor. Plast Reconstr Surg 117:565, 2006 17. Noser H, Hammer B, Kamer L: A method for assessing 3D shape
727 8. Kozakiewicz M, Elagal M, Walkowiak B, et al: Technical variation of fuzzy regions and its application on human bony or- 783
728 concept of patient-specific, ultrahigh molecular weight poly- bits. J Digit Imaging 23:422, 2010 784

FLA 5.4.0 DTD
YJOMS57738_proof
26 April 2017
2:44 pm
CE AH

 1.e8 TEMPLATE FOR ORBITAL WALL FRACTURE TREATMENT

785 18. Gunz P, Mitteroecker P: Semilandmarks: A method for quanti- 23. De Silva DJ, Rose GE: Orbital blowout fractures and race. 799
fying curves and surfaces. Hystrix 24:103, 2013 Ophthalmology 118:1677, 2011
786 800
19. Mitteroecker P, Gunz P: Advances in geometric morphometrics. 24. Lim CGT, Campbell DI, Cook N, et al: A case series of rapid pro-
787 Evol Biol 36:235, 2009 totyping and intraoperative imaging in orbital reconstruction. 801
788 20. Von Wilmowsky C, Bergauer B, Nkenke E, et al: A new, highly Craniomaxillofac Trauma Reconstruct 8:105, 2015 802
precise measurement technology for the in vitro evaluation of 25. Tabakovic SZ, Konstantinovic VS, Radosavljevic R, et al: Applica-
789 tion of computer-aided designing and rapid prototyping
803
the accuracy of digital imaging data. J Craniomaxillofac Surg
790 43:1335, 2015 technologies in reconstruction of blowout fractures of the 804
791 21. Szymor P, Kozakiewicz M, Olszewski R: Accuracy of open- orbital floor. J Craniofac Surg 26:1558, 2015 805
source software segmentation and paper-based printed 26. Podolsky DJ, Mainprize JG, Edwards GP, et al: Patient-specific
792 orbital implants: Development and implementation of technol-
806
three-dimensional models. J Craniomaxillofac Surg 44:202,
793 2016 ogy for more accurate orbital reconstruction. J Craniofac Surg 807
794 22. Dean D, Bookstein FL, Koneru S, et al: Average African American 27:131, 2016 808
three-dimensional computed tomography skull images: The 27. Metzger MC, Schoen R, Zizelmann C, et al: Semiautomatic pro-
795 cedure for individual preforming of titanium meshes for orbital
809
potential clinical importance of ethnicity and sex. J Craniofac
796 Surg 9:348, 1998 fractures. Plast Reconstr Surg 119:969, 2007 810
797 811
798 812

FLA 5.4.0 DTD
YJOMS57738_proof
26 April 2017
2:44 pm
CE AH