1 TITLE: 3-D PRINTING IN ORTHOPEDIC ONCOLOGY – WORK FLOW AND OUTCOMES

2 OF 59 CASES.

4 Abstract

6 Background: 3D printing has become an integral part of orthopedic oncology. Yet the penetration

7 of this technology is very less owing to lack of understanding of work flow or early failure due to

8 variables that wasn’t considered during the planning stage.

9 Aim: To report preliminary results, pitfalls and describe a workflow based on our experience

10 Methods: Descriptive, observation study of 59 cases done by assistance of 3D printing from

11 march 2016 to September 2021. An account of basic work flow, tips and pitfalls in planning and

12 institutional protocols are described. We categorized and analyzed cases based on clinical

13 parameters such as resection margin, operation time, blood loss. Number of cases that had

14 planning pitfalls also has been identified.

15 Results: resection via 3D printed jigs resulted in an average resection margin of 1.2mm,

16 operating time was lesser by an average of 25min [p-value:0.049] for 3D printed implants,

17 though there was no significant difference in blood loss [p-value: 0.24]. 3D planned aids had to

18 be abandoned in 3 cases due to unforeseen intraoperative challenges. In lower limb 3D printed

19 plates, average time to union was 4 months. There was no cases with non-union or delayed

20 union. Fluroscopy exposure was reduced significantly.

21 Conclusion: 3D printing assisted resection keeps resection to a minimum by giving adequate

22 oncological clearance. It also helps in reducing the operating time.

23

24 Introduction

25 Once ‘avant-garde’ in the field of healthcare, 3 Dimensional printing (3DP) has evolved from

26 cutting edge technology to a basic need in the management of several intricate conditions in the

27 field of surgery. 3DP helps in simplifying the byzantine cases in the field of surgery, and thereby

28 making mountains seem smaller for the reconstructive surgeons operating in complex territories.

29 The literature is not scanty on why 3DP should be in the kitty bag of every reconstructive

30 surgeon fighting the good fight. But the penetration of this must needed tool is very low in

31 practice because of the perplexity that the technology brings to the surgeons in general. The aim

32 of this paper is to disentangle and simplify the entire process for using 3D printing so that

33 orthopedic onco-surgeons can explore the full potential of this novel technology. We will present

34 a practical workflow, tips and tricks and technical pearls for using 3DP technology for

35 orthopedic oncology patients based on our experience of 180 cases spanning a decade.

36 Spectrum of application

37 1. 3D Anatomy Models

38 Complex surgeries are mostly termed so because of the complex anatomy of bone and soft tissue

39 at the area. A large pelvic tumor [pic] or Spine chordoma [pic] is a challenge to visualize as a

40 whole even from the most advanced scans and 3D reconstruction models on screen. Many times

41 a 3D reconstructed bone model superimposed with a tumor, sometimes even the vascular

2
42 structures (Figure ) gives an overview of the challenge and brings out more efficient and

43 confident dissections(1)

44

45 Figure: A shaft of femur osteosarcoma showing relation to femoral artery and branches to

46 the tumor. This branch was embolized preoperatively. Bone Cuts for ECRT are planned

47 pre-operatively on the model keeping adequate clearance.

48 In our experience these reduce the operative timing as the surgeon is well aware of the obscure

49 anatomy a large tumor has caused. The tortuous turns of the neurovascular structures may be

50 difficult to keep track of in relation to the anatomy even on contrast enhanced scans(2).

51 Tiptoeing around a huge tumor anticipating neurovascular structures can consume a lot of

52 operating time.

53 From a patient perspective, seeing a 3D model can enhance their understanding of the

54 procedure(3). We have found in our practice that patients become well versed with the

3
55 challenges the surgeon is going to face intraoperatively, and hence are more empathetic to a

56 postoperative outcome or a complication.

57 2. Patient Specific Cutting Guides

58 Taking cuts in a bone in orthopedic oncology is of utmost importance. While attempting to cut

59 lesser bone, the margin clearance can never be compromised. Even though the teaching is to err

60 on the bone stock and not in the margin clearance, the advent of technology gives us no excuse

61 to err here nor there. The classical approach of taking bone cuts by using the ruler that was

62 measured on MRI scans in an uneven bone is just a recipe for disaster. The jigs give the surgeon

63 the confidence of margin clearance by resecting the least amount of bone (Figure).

64

65 Figure: Shaft of Tibia OFD resected with tailor made cutting guide that could be secured

66 with k wires to preserve maximum bone.

67 Intraoperative reconciliation of bone cut using image intensification doesn't account for marrow

68 edema which might be a reactive zone of a malignant tumor. A repeat frozen section on a revised

69 bone cut increases the operative time by at least 20 minutes among other drawbacks that haunts a

70 frozen section confirmation(4).

4
71 Patient specific jigs have been employed by us for bone cuts in complex territories such a

72 pelvis(pic) and spine(pic). In pelvic tumor surgery, the complexity of the pelvic anatomy makes

73 a bony resection with the particular desired orientation for accurate fitting of the custom

74 prosthesis a challenge. In such areas, the depth controlled jigs that have been calibrated to the

75 saw we use, have limited our cuts to just the bone and thereby protect the vital structures

76 underneath. Injury to rectum, genitourinary and neurovascular during sacral chordoma excision

77 has been reported (5). Jigs have also given us confidence to cut the bone, without fretting under

78 the uni-dimensional image intensifier guidance as in case of a pelvic resection, where we try to

79 save the pelvic ring or leave behind adequate bone stock for prosthetic reconstruction. (pic). A

80 high degree of accuracy in resection of the bone is of utmost importance, given that the

81 prosthesis is made to 1:1 dimensions. Any errors in bone cuts may eliminate treatment options or

82 may result in inadequate resection margins. Cartiaux et al. (2008) demonstrated that 4

83 experienced surgeons could achieve a 10-mm resection margin, with 5-mm tolerance, on pelvic

84 sawbones in only half of the resections(6) . In juxta-articular lesions, the cutting guides may be

85 instrumental in saving the joint or the physis.

86 3. 3D printed implants

87 Once recycled autograft was increasingly used in limb salvage surgery, the need of the hour was

88 patient specific implants. Tumor resection margin dicated where screws can be put in to refix the

89 bone that had undergone ECRT or pedicled liquid nitrogen freezing. Standard plate designs used

90 for trauma were largely useless in this cohort of patients. Availability of pediatric plates in all

91 sizes was a difficulty. This is a glaring problem because a large portion of the recycled autograft

92 surgeries were done in the pediatric age.group.(Figure)

5
 93 Custom made plates were printed in titanium with provision for locking screws. Such plates

94 could be printed as per the profile, length, number of screws, direction of screws and deformity

95 in the bone. Locking screws with bone specific lengths as measured on MRI were printed and

96 tried on a 3D printed model of the same patient preoperatively. This ensured there were less

97 errors intraoperatively and procedure was a lot quicker.

98

99 Figure - Distal femur OS with involvement of physis as well. A cut distal

100 to physis was only way to save the joint. Here the joint was saved due to a

101 reliable distal femur plate with custom designed screw direction and

6
102 number. Plate is well contoured to the pediatric femur and is holding

103 recycled autograft and vascularised fibula.

104 4. 3D printed prosthesis

105 Standard endoprosthesis that is readily available in the market covers most of the mega-

106 reconstruction needs of a tumor surgeon. Prosthesis of extremities are rarely printed(7)]. In our

107 practice massive endoprosthesis printing was done for pelvic and shoulder girdle tumors. The

108 deficit created by the resection was planned in advance and the prosthesis was designed with

109 computer assisted drawing.

110

111 Figure: Shoulder Chondrosarcoma which could be managed by limb salvage because of

112 this custom made 3D printed prosthesis. A reverse shoulder prosthesis was implanted into

113 this printed prosthesis for the articulation.

7
114 Point of fixation was planned with screws, plate-screw, posts, cement, and interlocking bolts.

115 Prosthesis was coated with hydroxyapatite for integration to host bone or allograft. Provision was

116 added for attachment of muscles, tendons and ligaments.

117 METHODOLOGY

118 This paper has been written based on our experience as a comprehensive and an exclusive cancer

119 centre in the third most populous city in India. The department of Orthopedic Oncology at our

120 centre sees over 1100 patients and operates close to 150 patients annually. Our practice with

121 respect to 3D printing has gone through abundant transitions over the years as the costs of

122 technology changed. We present an descriptive, observational, retrospective, and monocentric

123 study of 59 cases assisted by 3DP technology between 2016 to 2021. [Chart 1]

MRI image aquisition,

CT image aquisition segmentation and Planning resection
and segmentation superimposition on CT planes
images

implant/prothesis
Jig planning Printing PLA model
planning

Final printing and post

Dry runs
processing

124

8
125 1. FINDING THE APPROPRIATE INDUSTRIAL SUPPORT

126 There is no dearth for companies that offer 3D printing support across the world. Delay or

127 inexpertise can erode away valuable waiting period for a cancer patient and cause delays which

128 can result in less satisfactory outcomes. A team should be evaluated in the following key steps.

129 1.1 Segmentation

130 This is a process in which the bone and tumor or structure in question, such as nerve, vessels etc

131 are delineated from the CT/MRI scans. These can be done on CAD [Computer Aided Design]

132 softwares. The process of segmentation has evolved from being completely manual to semi

133 automatic to fully automatic. Manual segmentation will require step by step drawing of the tumor

134 and bone in every slice. It's a time consuming process. Semi automated and fully automated

135 systems on the other hand is a much faster process where an image is divided into regions with

136 similar properties such as gray level, color, texture, brightness, and contrast. A fully automated

137 system has a high failure rate owing to the complex nature of the tumor. The output of the

138 segmentation algorithm is affected due to imaging issues like partial volume effect, presence of

139 artifacts, intensity inhomogeneity, closeness in gray level of different soft tissue etc (8). We

140 follow a semi automated system in which borders of a software segmented tumor or anatomical

141 structure is revised by engineer and radiologist in conjunction to get maximum accuracy. In

142 summary it's imperative that the biomedical engineers that one associates with, have a semi

143 automated segmentation proficiency. In such a system, segmentation wouldn't take more than an

144 hour.

145 1.2 Printing

9
146 It’s important to have an idea of what printing technique the industry that we employ is using.

147 There are 4 types of printing techniques that’s most common. Appropriate technique is used as

148 per our requirement.

149 Sterolithography [SLA] printing can be used to generate bone models. It works by melting resins

150 hence gives a better finish to the models.

151 Fused deposition modeling [FDM] uses plastic polymers. [polylactic acid – PLA, Acrylonitrile

152 Butadiene Styrene- ABS]. It prints by layered deposition and so the overall finish of the surface

153 is of lesser quality but its cost effective owing the comparatively cheaper price of the printer and

154 raw material compared to SLA, and hence these are most commonly used by non-professionals.

155 Laser sintering allows use of metals and other materials for printing. Selective laser sintering

156 [SLS] can be used for non-metal printing purposes. Direct metal laser sintering [DMLS] is suited

157 for metal printing and for obvious reasons is the most expensive technique to execute. These are

158 mostly used in industrial grade printing. Such printers are used for prosthesis and implant

159 printing.

160

161

162 1.3 Preoperative planning - Tips and Pitfalls

163 1.3.1 Bone

164 When measuring bone cuts, we have to keep in mind that there is a chance for the measurement

165 to be under presented in CT/MRI. Hence when planning a bone resection we tend to take an extra

10
166 centimeter for margin more than what is measured in the MRI/CT. 1:1 models can be smaller

167 than the actual model.

168 1.3.2 Soft tissue

169 Soft tissue obstruction to seating an implant can happen in the following situations, hence has to

170 be carefully checked for before planning a jig or implant,

171 a. Radiation induced soft tissue contracture.

172 b. In delayed reconstruction, long standing deformity may have caused ligamentous

173 laxity/contracture.

174 c. Soft tissue bulk at a bony site may bar snug fit of the jig onto the bony prominence. Hence

175 another site of jig fixation may have to be planned. [Figure Abhijeet]

176 1.3.3 Neurovascular structures

177 Neurovascular structures can get grossly displaced after removal of a large tumor. These may

178 have to be released or reconstructed to ensure adequate seating of the implant or prosthesis.

179

180

181 1.4 Communications to the radiology desk

182

183 1.4.1 CT image aquisition

11
184 This protocol describes the guidelines followed by us for a CT scan for ordering Titanium 3D

185 Printed Patient Specific Implants, Plates and Guides and anatomic models.

186 1.4.1.1 General Considerations

187 a. CBCT Scans are not accepted for PSI.

188 b. Patient specific devices will be designed to fit the patient anatomy at the time of the CT scan.

189 c. Changes in the patient anatomy occurring after the CT scan, as well as the use of the device

190 after such changes, may result in a suboptimal fit of the device or implant. Scans must be less

191 than 2 months old. We use the following scan parameters or the closest approximation possible.

192 1.4.1.2 Preparation of the patient for scan:

193 a. Remove any non-fixed metal prosthesis or jewelry.

194 b. Non-metal dentures may be worn during the scan.

195 c. Make the patient comfortable and instruct him not to move during the procedure.

196 d. Normal breathing is acceptable but any other movement, such as tilting and/or turning the

197 head, can cause motion artifacts that compromise the reconstructed images requiring the

198 patient to be rescanned.

199 e. Stabilize the relationship of the jaws during the scan. The patient is preferably scanned

200 with a very thin bite wafer that does not influence the facial soft tissues.

201 1.4.1.3 Patient Positioning:

202 a. Place the patient supine on the scanner table and move the patient into the gantry,

203 headfirst.

12
204 b. Minimize the artefacts caused by metallic dental restorations or orthodontic brackets by

205 aligning the patient's

206 c. occlusal plane as much as possible with the axial slices.

207 d. Do not deform the soft tissue.5 . Depending on the product or service requested, the field

208 of view should include:- Nose and chin, Left and right TMJ, Other regions of interest if

209 required (ex. cranium), For reconstruction cases the complete tumour/defect. Region of

210 interest should be covered in the Field of view.

211 1.4.1.4 CT Scanning Instructions

212 a. Images scanned under a gantry tilt and oblique or reformatted images negatively

213 influence the accuracy, its recommended only primary axial images.

214 b. All slices must have the same field of view, reconstruction centre, and table height.

215 c. Scan with the same slice spacing, less than or equal to the slice thickness.

216 1.4.1.5 Reconstruction of the images CT

217 Use these listed CT Scan Parameters for image reconstruction or as closest approximation

218 possible:

219 a. Gantry tilt/oblique angle 0°

220 b. Matrix 512x512

221 c. Slice thickness Maximum 1,0 mm

222 d. Feed per rotation Maximum 1.0 mm

223 e. Reconstructed slice increment Maximum 1.0 mm.

224 f. Pixel size or Pixel spacing must be less than 0.5 mm.

13
225 Accepted media: Standard DICOM format- CD or DVD.

226

227 1.4.2 MRI acquisition protocol

228 This protocol describes the guidelines used by us for an MRI scan that is taken for the purpose of

229 ordering a 3D printed anatomical model and implants. This protocol is preferably transferred to

230 the radiology department, together with the scan order. Using this scanning protocol will result in

231 a more accurate model.

232 1.4.2.1 Preparation of the patient

233 a. Remove any non-fixed metal dentures or prosthesis, in addition to any jewelry that might

234 interfere with the region to be scanned. Place the patient supine on the scanner table and

235 move the patient into the gantry, headfirst

236 b. Make the patient comfortable and instruct him not to move during the procedure. Normal

237 breathing is acceptable, but any other movement, such as tilting and turning the head can

238 cause motion artifacts that compromise the reformatted images, requiring the patient to

239 be rescanned.

240 1.4.2.2 Aligning the patient

241 a. It is very important to know whether a patient has been scanned with a gantry tilt

242 lan angulation other than 01 or not Although the software has been adapted to

243 support data scanned with gantry tilt, interpolations and 3D representation will

244 have an inferior quality due to the gantry tilt.It is therefore advised not to use a

245 gantry tilt.

14
246 b. If the only option is to use a gantry tilt, please indicate the direction of angulation

247 when the data is sent. The direction of angulation:

248

249

250

251 c. Align the patient in a way that prevents as many artifacts as possible in the resulting

252 images.

253 d. Use the head holder with sponges to stabilize the position. If you cannot orient the head

254 properly in the head holder, use the tabletop. In either case, strap the head securely to

255 prohibit motion.

256 e. Stabilize the relationship of the jaws during the scan. The patient is preferably scanned

257 with the jaws slightly open lif available, you can use a bite block. This will reduce the

15
258 risk of artifacts from the opposing jaw disturbing the images of the jaw of interest. Also,

259 this will make it possible to isolate the occlusal plane from the images.

260 f. You can take a lateral alignment image (called a Localizer, Scout view,

261 topogram,Scanogram, Pilot or Sunview depending on the MR manufacturer] to verify

262 the correct patient positioning.

263 1.4.2.3 Scanning instructions

264 a. Set the table height so that the area that needs to be scanned is centred in the scan

265 field.

266 b. All slices must have the same field of view, the same reconstruction centre, and

267 the same table height.

268 c. Not overlapping the axial slices can reduce the quality of the reformatted images.

269 d. Scan all slices of the study in the same direction.

270 e. Scan with the same slice spacing; the slice spacing must be less than or equal to

271 the slice thickness.

272 f. The slice thickness should preferably not be larger than 1 mm.

273 g. Scan Direction: Axial or Sagittal Matrix [Preferred]: 512x512, Slice thickness

274 (Preferred): Less than 1 mm

275

16
276

277

278 1.4.2.4 Image Reconstruction:

279 Use a proper image reconstruction algorithm to get sharp reformatted images where you can

280 locate internal structures such as the alveolar nerve and for bony anatomy use bone or high-

281 resolution algorithm.

282 Accepted media: Standard DICOM format- CD or DVD.

283

17
284 1.5 Communications to the Engineer

285 a. Incision area and bony landmarks that can be used for jig positioning or implanting a

286 prosthesis has to be marked out.

287 b. large prosthesis should be designed with a lattice structure to bring down the weight of the

288 prosthesis.

289 c. For prosthesis with large working length, an additional screw fixation apart from cementing

290 should be planned

291 d. Length and breadth of the saw used has to be communicated for efficient planning of jigs so

292 that a stop can be planned on the jig to avoid any inadvertent injury to vital structures.

293 e. Jigs should be planned with provision for k-wire fixation.

294

295 1.6 Preoperative dry runs.

296 a. Check for sagittal plane mismatch

297 b. Confirm all screw holes are directed properly

298 c. plan screw length in advance

299 d.

300

301 Results

18
302 59 patients underwent planning and management with 3D printing technology. All patients had

303 patients had anatomy model printed. 25 patients had jigs planned. 16 patients had 3D printed

304 implant used, 6 patients had 3D printed prosthesis planned.

305 56 patients underwent the planned procedure. One patient planned for 3D printed pelvic

306 prothesis had to abandoned intraoperatively after encountering extensive soft tissue contracture

307 in the pelvis from adjuvant radiation, This was not taken into consideration in the preoperative

308 planning. One pediatric patient with osteofibrous dysplasia of the tibia was planned for resection

309 of the segment with 3D printed jigs. Since the tibialis anterior insertion and thick peristoeum

310 wasn’t taken into consideration, the jig wouldnt fit snugly onto the anterior surface of the tibia.

311 The resection had to be carried out under C arm guidance after further exposure. Another patient

312 planned for 3D printed plate in a case of pedicled freezed distal femur osteosarcoma had to be

313 done using conventional plates because logistical delay in planning and printing. The

314 conventional plate was illfitting as the plate was proud at the greater trochanter.

315 All cases had R-0 resection with average resection margin being 23mm [range 10mm-38mm].

316 Frozen section was done intraoperatively for all patients.

317 In 3 cases were jigs were used for posterior bone cut for type 1/4 resections in pelvis reduced the

318 surgical time by 35 minutes average in the 3 cases compared to 3 other similar cases were jigs

319 weren’t used. Additional time was required to do more exposure and to place careful cuts so as to

320 not injure the structures behind. In 8 cases of femur resections, time difference didn’t reach

321 significance.

322 There was no significant difference [p-value: 0.24] in average blood loss when compared to

323 similar anatomical sites operated without 3D printed jigs.

19
324 In 12 cases of 3D printed plates for femur resections followed by recycled autograft [n=8] or

325 massive allograft reconstruction [n=4], the average time for implantation of the graft + free

326 fibula construct was 18 minutes whereas using conventional plates average time was 38 minutes.

327 Fluroscopy was used only for final screening of the implant fixation. In a case of conventional

328 plating, mean exposure time was 4.09 minutes and 0.1-0.22 mSv radiation was emitted.(9)

329 Average follow-up is 28 months [range of 10 - 36 months]. Average time to union in cases

330 operated with 3D printed plate construct in lower limb was 4 months [Range of 3 – 6 months].

331 There was no instability, looseing or periprosthetic fracture in this series of cases with 3D printed

332 prosthesis.

333

334 Discussion

335 3D printing aids gives the advantage to reduce operating time, reduce the need for frequent

336 fluroscopy exposure, avoid implant malpositioning, decrease intraoperative blood loss. 3D

337 printed anatomical models aids the surgeon’s 3D orientation of the anatomy which can improve

338 the flow of surgical care. Oncological surgical treatment is a trade off between adequate margins

339 and function, with margin being more important. Rough calculations on radiological imaging

340 without proper landmarks to depend on during the surgery makes bone resections dangerous.

341 Resections may vary from going through the tumor increasing the chances of spillage, to

342 resections far away taking more normal bone than required.

20
343 3D printed plates has provides snug fitting implant for pediatric population, were most standard

344 plates are illfitting. We have had 2 patients with delayed union which was managed by bone

345 grafting at 9 months post op compared to no nonunions in the 3D printed plate group.

346 One of the major drawbacks is the time taken for execution. In a case of distal femur

347 osteosarcoma planned for 3D printed plate, we had to abandon the planning at a stage because

348 the delay in 3D printed would have adversely affected the patient outcome. The collaboration of

349 the surgeon, radiologist and the engineer becomes another bottleneck in the logistics.

350 Should you buy a 3D printer?

351 Having a 3 D printer as in-house requires a very efficient work flow if it has to be economically

352 viable. Challenges are peasant riddled in every step of the way from owning, designing and

353 printing models and implants.

354 In India, one might need an ISO certification of standardization to own and print medical

355 implants. Most countries need approval from their standardization agencies, FDA for the US and

356 European Union medical device regulation for all the countries in the EU.

357 The logistics for owning an 3D printer starts with having a space that has sufficient aeration and

358 temperature control. Workforce includes biomedical engineers,designers specialised in computer

359 assisted drawing [CAD]. Purchasing and stocking the raw material can run into problems

360 particularly if the materials have to be imported. Global goods movement restrictions owing to

361 developments in the geopolitical climate and novel covid coronavirus infections have been a

362 logistic nightmare for many centers. Technology for designing and printing is fast progressing

21
363 bringing down the cost and the time taken for processing. Keeping up with the upgradation in the

364 softwares for designing and hardware of printers can be an expensive affair.

365 All in all, a large volume of 3D printing requirement, a well run interdisciplinary workflow

366 between surgeons, radiologists, biomedical Engineers and other technician assistants, a hospital

367 management who is patient and resourceful to accept the overwhelming financial investment, is

368 paramount in running a successful in-house 3D printer.

369 Our current practice has a team of surgeons doing the planning, and a radiologist doing the

370 segmentation. We have liaised with an industrial 3D printing company who does the designing

371 and the printing of implants and prosthesis. We have recently acquired a low cost FDM printer

372 which uses PLA polymer as raw material. It’s mainly used for generating bone models for patient

373 education and surgical orientation.

374

375 Costing

376 Costing depends on the requirement of the modality. Basic anatomy models will cost ranging

377 from 100$ - 200$ depending on the size of the specimen being printed. Although the printing of

378 the jig maybe an inexpensive affair, the cost involves the time spend by the team on developing

379 and planning the jig. 3D printed implants such a titanium plate would cost 800$-1000$.

380 Prosthetic printing is an uphill task in terms of planning and printing. Larger amount of raw

381 material will be required and it may take more than 60 hours to print. Average costs incurred for

382 a pelvic reconstruction implant was 4000$.

383 Conflict of interest; None

22
384 Funding: None
385

23
386 Conclusion

387

388 3D printed aids, viz anatomy models, patient specific cutting jigs, implants and prosthesis

389 represents a need of the hour in the orthopedic oncology especially in pedaitric cases and area of

390 complex anatomy like pelvis. Even though it is overshadowed by the complexity and time

391 consumed for planning , interdepartmental collaboration, costing, and logistics, it can improve

392 resections, decrease time of surgery, reduce radiation exposure, and give better long term

393 outcomes.

394

24
395 Bibliography

396

397 1. Tam MD, Laycock SD, Bell D, Chojnowski A. 3-D printout of a DICOM file to aid surgical

398 planning in a 6 year old patient with a large scapular osteochondroma complicating

399 congenital diaphyseal aclasia. J Radiol Case Rep 2012;6:31-7.

400 2. Chang, YC.C., Ackerstaff, E., Tschudi, Y. et al. Delineation of Tumor Habitats based on

401 Dynamic Contrast Enhanced MRI. Sci Rep 7, 9746 (2017). https://doi.org/10.1038/s41598-

402 017-09932-5.

403 3. D’Urso PS, Redmonda MJ. A method for the resection of cranial tumours and skull

404 reconstruction. Br J Neurosurg 2000;14-16.

405 4. Jaafar H. Intra-operative frozen section consultation: concepts, applications and limitations.

406 Malays J Med Sci. 2006;13(1):4-12.

407 5. Peter Paul Varga, Zsolt Szövérfi & Aron Lazary (2014) Surgical treatment of primary

408 malignant tumors of the sacrum, Neurological Research, 36:6, 577-587, DOI:

409 10.1179/1743132814Y.0000000366.

410 6. Cartiaux O, Docquier P, Paul L, Francq BG, Cornu OH, Delloye C, Raucent B, Dehez B,

411 Banse X. Surgical inaccuracy of tumor resection and reconstruction within the pelvis: An

412 experimental study. Acta Orthop 2008; 79 (5): 695-702.

413 7. Three-dimensional printed calcaneal prosthesis following total calcanectomy.Imanishi J,

414 Choong PFInt J Surg Case Rep. 2015; 0():83-7.

25
415 8. Sharma N, Aggarwal LM. Automated medical image segmentation techniques. J Med Phys.

416 2010;35(1):3-14. doi:10.4103/0971-6203.58777.

417 9. Mahajan A, Samuel S, Saran AK, Mahajan MK, Mam MK. Occupational radiation exposure

418 from C arm fluoroscopy during common orthopaedic surgical procedures and its prevention.

419 J Clin Diagn Res. 2015;9(3):RC01-RC4. doi:10.7860/JCDR/2015/10520.5672.

420

26