Original Article

Optical Topographic Imaging for Spinal Intraoperative Three-Dimensional Navigation

in Mini-Open Approaches: A Prospective Cohort Study of Initial Preclinical and

Clinical Feasibility

Daipayan Guha1-3, Raphael Jakubovic3,4, Naif M. Alotaibi1,2, Jesse M. Klostranec5, Sidharth Saini5, Ryan Deorajh3,

Shaurya Gupta3,6, Michael G. Fehlings1,2, Todd G. Mainprize1, Albert Yee7, Victor X.D. Yang1-3,8

-OBJECTIVE: Computer-assisted three-dimensional navi-gation often guides spinal instrumentation. Optical topo-graphic imaging (OTI) offers comparable accuracy andsignificantly faster registration relative to current naviga-tion systems in open posterior thoracolumbar exposures.We validate the usefulness and accuracy of OTI in mini-mally invasive spinal approaches.

-METHODS: Mini-open midline posterior exposures wereperformed in 4 human cadavers. Square exposures of 25, 30,35, and 40 mm were registered to preoperative computedtomography imaging. Screw tracts were fashioned using atracked awl and probe with instrumentation placed. Navi-gation data were compared with screw positions on post-operative computed tomography imaging, and absolutetranslational and angular deviations were computed.

In vivo validation was performed in 8 patients, withmini-open thoracolumbar exposures and percutaneousplacement of navigated instrumentation. Navigated instru-mentation was performed in the previously describedmanner.

-RESULTS: For 37 cadaveric screws, absolute trans-lational errors were (1.79 � 1.43 mm) and (1.81 � 1.51 mm)in the axial and sagittal planes, respectively. Absoluteangular deviations were (3.81 � 2.91�) and (3.45 � 2.82�),respectively (mean � standard deviation). The number of

Key words- Computer-assisted surgery- Image guidance- Minimally invasive surgery- Navigation

Abbreviations and Acronyms3D: Three-dimensionalCAN: Computer-assisted navigationCT: Computed tomographyMIS: Minimally invasive surgeryOTI: Optical topographic imagingSD: Standard deviation

From the 1Division of Neurosurgery, Department of Surgery; 2Institute of Medical Science,School of Graduate Studies, University of Toronto, Toronto, Ontario; 3Biophotonics and

WORLD NEUROSURGERY 125: e863-e872, MAY 2019

surface points registered by the navigation system, but notexposure size, correlated positively with the likelihood ofsuccessful registration (odds ratio, 1.02; 95% confidenceinterval, 1.009e1.024; P < 0.001).Fifty-five in vivo thoracolumbar pedicle screws were

analyzed. Overall (mean � standard deviation) axial andsagittal translational errors were (1.79 � 1.41 mm) and (2.68� 2.26 mm), respectively. Axial and sagittal angular errorswere (3.63� � 2.92�) and (4.65� � 3.36�), respectively. Therewere no radiographic breaches >2 mm or any neuro-vascular complications.

-CONCLUSIONS: OTI is a novel navigation techniquepreviously validated for open posterior exposures and inthis study has comparable accuracy for mini-open mini-mally invasive surgery exposures. The likelihood of suc-cessful registration is affected more by the geometry of theexposure than by its size.

INTRODUCTION

ntraoperative three-dimensional (3D) computer-assisted nav-igation (CAN) has become standard of care in cranial

Ineurosurgery for the localization of subsurface anatomy.

Spinal CAN often guides instrumentation placement and tissue

Bioengineering Laboratory, Sunnybrook Health Sciences Centre, Toronto, Ontario;4Departments of Biomedical Physics, Ryerson University, Toronto, Ontario; 5Department ofMedical Imaging, University of Toronto, Toronto, Ontario; 6Faculty of Applied Sciences andEngineering; 7Division of Orthopedic Surgery, Department of Surgery, University of Toronto,Toronto, Ontario; 8Department of Electrical and Computer Engineering, Ryerson University,Toronto, Ontario, Canada

To whom correspondence should be addressed: Daipayan Guha, M.D. Ph.D.[E-mail: deep.guha@mail.utoronto.ca]

Citation: World Neurosurg. (2019) 125:e863-e872.https://doi.org/10.1016/j.wneu.2019.01.201

Journal homepage: www.journals.elsevier.com/world-neurosurgery

Available online: www.sciencedirect.com

1878-8750/$ - see front matter ª 2019 Elsevier Inc. All rights reserved.

www.journals.elsevier.com/world-neurosurgery e863

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

resection; however, adoption has been limited by cumbersomeand lengthy registration protocols, workflow hindrances, steeplearning curves, and high costs.1-6

The usefulness of CAN is most apparent in minimally invasivesurgery (MIS) and deformity-correcting procedures, in whichanatomic landmarks are not directly visible or are significantlydistorted.1,5,7,8 MIS techniques, through mini-open, tubular, and/or endoscopic approaches, have been shown to shorten hospitallength of stay, minimize intraoperative blood loss, and improveshort-term patient-reported outcomes. The impact on operativetime and postoperative complications, relative to comparable openspinal procedures, remains to be defined.9-14 However, MIS ap-proaches have typically been guided by intraoperative fluoroscopyor computed tomography (CT). These techniques are associatedwith substantial radiation exposure and workflow disruption.15

Optical topographic imaging (OTI) is a novel technique for 3Dsurface acquisition, patient-to-image registration, and intra-operative navigation, developed by our research group. OTI reg-isters significantly faster than CAN systems with comparableaccuracy and without intraoperative radiation exposure.16 This

Video available atwww.sciencedirect.com

technology obviates many of the limitations of CANtechniques.1,5 In its current iteration, OTI requiresline of sight to exposed bony anatomy to allowmachine-vision cameras to generate a virtual 3D sur-face for patient-to-image registration. OTI has beenvalidated only in open posterior thoracolumbarapproaches with incisions exposing >3 spinal levels.In this study, we assess the ability of OTI to

Figure 1. Mini-open posterior midline exposure at T2, in preclinicalcadaveric validation. Dynamic reference frame for optical topographicimaging navigation is clamped to the T2 spinous process. Exposure sizeof 25 � 25 mm has been simulated with sterile towels.

perform successful patient-to-image registration and accurateintraoperative navigation in mini-open spinal procedures. Weexplore predictors of successful registration and their correlationwith quantitative navigation accuracy.

METHODS

Reporting of all methodology is performed in accordance with thecriteria for STROBE (Strengthening the Reporting of Observa-tional Studies in Epidemiology [www.strobe-statement.org]).

Specimen/Patient SelectionPreclinical validation was performed in 4 human cadavers. All ca-davers underwent preoperative and postoperative helical CT imag-ing at 0.5 mm slice thickness. Institutional ethics board approvalwas obtained (institutional review board number 16-0051-E).For human in vivo clinical testing, 8 patients without history of

previous spinal surgery were enrolled in an ongoing trial of OTInavigation at Sunnybrook Health Sciences Centre (institutionalreview board numbers 309-2014 and 086-2015). Mini-open ap-proaches were chosen in these patients at the attending surgeon’sdiscretion, based on a combination of appropriate body habitus,minimal need for deformity correction, and minimal or no needfor focal decompression. All patients underwent preoperative andpostoperative helical CT imaging, reformatted at 0.625 mm slicethickness.

Surgical TechniqueCadavers were placed prone on a standard operating table, withhead fixed in a Mayfield clamp. Mini-open midline posterior

e864 www.SCIENCEDIRECT.com WORLD NE

exposures of the spinous process and medial bilateral hemi-laminae were performed at T2, T6, T10, and L3 (Figure 1). Abladed self-retaining retractor was placed. All cadaveric proced-ures were performed by a single surgeon (D.G.).In human in vivo clinical testing, posterior instrumented fusions

were performed for stabilization after either trauma or iatrogenicinstability after transpedicular decompression of bony and epiduraltumor. All patients were positioned prone on aWilson frame. Mini-openmidline exposures were performed for OTI registration as wellas laminectomy (Figure 2). Similar mini-open exposures to cadav-eric testing were performed, maintained by identical bladed self-retaining retractors. All in vivo procedures were performed by asingle surgeon (V.X.D.Y.), with trainee assistance.

Registration and Intraoperative NavigationIn cadaveric studies, the retractor width was increased serially tocreate square exposures of size 25� 25, 30� 30, 35 � 35, and 40 �40 mm (Figure 1). At each level, the exposed anatomy for eachexposure size was registered to preoperative CT using OTI.Technical details of OTI registration are described separately.16

UROSURG

A structured-light pattern is projected onto theexposed anatomy and recorded by stereoscopic cam-eras to reconstruct a 3D surface (Figure 3 and Video 1).This surface is automatically aligned to preoperativeCT imaging using a registration algorithm in realtime. In our previous validation of OTI for openthoracolumbar spinal procedures, registration timefor a single level was found to be 41 seconds.

Registration accuracy was verified manually by placing anoptically tracked awl on bony landmarks and assessing correlationto the navigation display. Registration was deemed successful ifthe OTI system captured sufficient anatomy for patient-to-imageregistration (�100 surface points) and if manual verification bythe operator showed acceptable accuracy using identifiableanatomic landmarks with visual and tactile feedback. The number

ERY, https://doi.org/10.1016/j.wneu.2019.01.201

Figure 2. Mini-open posterior midline exposure at T8-9for patient A (A), T10-11 for patient B (B), and L2 forpatient C (C) in clinical in vivo validation. Dynamicreference frame (arrowhead) is clamped to an exposed

spinous process in (A), (B) and (C). A tracked drill guide(arrow) and K-wires were used for percutaneousplacement of instrumentation (D).

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

and location of surface points used by the OTI system for regis-tration were also recorded. At each level, the 30-mm � 30-mmexposure was used to place instrumentation. A tracked awl andgearshift probe were used to fashion pedicle screw tracts at eachregistered level. Cortical trajectory tracts were also fashioned at L3.Titanium screws were placed at each level.In human in vivo studies, midline mini-open exposures were

used for OTI registration, with similar registration verification asin cadaveric specimens, especially using tactile feedback from atracked awl percutaneously on bony landmarks. Screw tracts werethen fashioned percutaneously using the tracked awl, gearshiftprobe, and/or drill guide, followed by placement of Kirschnerwires. Appropriately sized cannulated titanium pedicle screwswere placed percutaneously over the Kirschner wires using astandard untracked screwdriver (Figure 2).

WORLD NEUROSURGERY 125: e863-e872, MAY 2019

Evaluation of Navigation AccuracyAbsolute quantitative navigation accuracy was measured bycomparing the final screw position on postoperative CT imagingwith a screenshot of the planned screw trajectory on the navigationsystem intraoperatively. Translational and angular deviations fromthe planned entry point and trajectory were quantified in the axialand sagittal planes using multiplanar reformatting of both pre-operative and postoperative CT imaging. The method of absolutenavigation error quantification has been described by our grouppreviously (Figure 4).17-19

Radiographic accuracies of all in vivo screws were gradedindependently by 2 radiologists (J.M.K., S.S.), using both the2-mm and Heary classifications.20,21 Screws were dichotomized asacceptable (2 mm grade �2; Heary grade �2) or poor (2-mm grade>2; Heary grade >2) per convention.17,20

www.journals.elsevier.com/world-neurosurgery e865

Figure 3. Computer-assisted design model of optical topographicimaging navigation unit integrated into surgical light head.Structured-light projector (arrow), stereoscopic cameras forthree-dimensional surface mapping (arrowheads) and infrared camerasfor tool tracking (*) are shown.

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

All image processing and measurements were performed usingan OsiriX 64-bit workstation (version 10.9.5 [Pixmeo SARL,Geneva, Switzerland]).

Statistical AnalysisDifferences in absolute navigation errors between spinal levelswere quantified with 1-way analysis of variance, with a Tukeyhonest-significant-difference test for post hoc comparisons. Cor-relation between the likelihood of successful registration and thenumber of surface points used for patient-to-image registration aswell as the size of the exposed anatomy was calculated usingmultiple logistic regression models. Hierarchical mixed-effectsgeneral linear modeling was used to adjust for second-order dif-ferences between cadavers/patients when required based on uni-variate analyses. In vivo MIS cases were matched 1:2 based on age/gender/spinal level and separately based on mean pedicle diameterto patients who had undergone open posterior thoracolumbarinstrumentation using OTI guidance in our previous trial.16 Themethods for our open trial are described separately in Ref.16;however, patient positioning and OTI setup were identical tothose used for patients undergoing in vivo MIS in this study.Significance levels for all tests were set at a < 0.05.All statistical analyses were performed in SPSS version 21 (IBM

Corp., Armonk, New York, USA).

RESULTS

For the 4 cadavers used in preclinical validation, mean age atdeath was 91.4 years (range, 83e96 years). Thirty-seven screwsfrom the 4 cadavers were included in our analysis: 8 pedicle screwsat T2, 10 at T6, 9 at T10, and 4 pedicle and 6 cortical screws at L3.One pedicle at T10 was not analyzed because of the unavailabilityof appropriate instrumentation to place at this level.

e866 www.SCIENCEDIRECT.com WORLD NE

In vivo clinical feasibility was assessed in 8 patients, with meanage 57.2 years. Fifty-five thoracolumbar pedicle screws placed withCAN from mini-open OTI registrations were analyzed, bilaterallyat T8-9 for patient A, T6-11 for patient B, T12-L4 for patient C,T12-L1 for patient D, T11-L3 for patient E, T10-L2 for patients Fand G, and T9-11 for patient H. Mean (�standard deviation [SD])pedicle diameter was 4.42 � 3.51 mm.

Patient-to-Image RegistrationIn the cadaveric study, we systematically studied the attributes ofsuccessful registrations with OTI and compared them with un-successful registrations. A total of 131 registration attempts weremade through mini-open exposures of varying sizes, with 71.8%verified by the operator as successful based on correlation betweenimaging and anatomic bony landmarks (Table 1). The likelihoodof successful registration was greater at T2 than at any othertested spinal level (odds ratio, 6.02; 95% confidence interval,1.47e24.63; P ¼ 0.013). The minimum tested exposure of 25 �25 mm allowed successful registration in 66.7% of attempts atT2, 80.0% of attempts at T6, 33.3% of attempts at T10, and25.0% of attempts at L3. Successful registration in >70% ofattempts necessitated a minimum exposure of 25 � 25 mm atT6, 30 � 30 mm at T2 and T10, and 35 � 35 mm at L3. Themean wound depths at T2 (5.89 cm) and L3 (5.68 cm) weresignificantly greater than at T6 (3.23 cm) or T10 (3.50 cm)(P < 0.001); however, wound depth did not correlate with thelikelihood of successful registration.Overall, (431 � 235) surface points (mean � SD) were used by

the OTI system for patient-to-image registration, (502 � 231)points for successful registrations, and (250 � 120) for unsuc-cessful registrations. Significantly fewer points were acquired andused by the system at the smallest exposure of 25 � 25 mm (mean,303 points; P ¼ 0.039), with no significant differences in thenumber of points registered at 30 � 30, 35 � 35, or 40 � 40 mmexposures (Figure 5).In multiple logistic regression modeling, the number of surface

points registered by the OTI system correlated positively with thelikelihood of successful registration, independent of spinal level,exposure size, and wound depth (odds ratio, 1.02; 95% confidenceinterval, 1.009e1.024; P < 0.001). Exposure size itself was notindependently associated with the likelihood of successful regis-tration (Table 1).In human clinical testing, 9 registrations through MIS expo-

sures were performed, 2 in patient B for the placement of T6-8 andT9-11 screws, respectively, and 1 registration each for all otherpatients. All registrations were successful on the first attempt,using the representative exposures shown in Figure 2. The meannumber of registered points per level was 752 � 186 (mean � SD).

Quantitative Navigation AccuracyIn cadaveric testing, overall (mean � SD) axial and sagittaltranslational errors were (1.79 � 1.43 mm) and (1.81 � 1.51 mm),and axial and sagittal angular errors were (3.81� � 2.91�) and(3.45� � 2.82�), respectively. There were no significant differencesin errors between levels or between pedicle and cortical trajectoryscrews (Figure 6). The number of points registered by OTI did notsignificantly correlate with any metric of absolute navigation error.

UROSURGERY, https://doi.org/10.1016/j.wneu.2019.01.201

Figure 4. Measurement of absolute navigationaccuracy, in the axial (A, C) and sagittal (B, D) planes.Comparison is made between intraoperativenavigation screenshots of planned entry points andtrajectories (A, B) and final screw placement onpostoperative computed tomography (C, D).Reference lines (dashed) are drawn, in the axial plane,

in the midsagittal line (bisecting the vertebral body,spinal canal, and spinous process) and, in the sagittalplane, along the inferior end plate. Translational error iscomputed as (d1 � d ); angular error is computed as(B1 �B). Image taken with permission from The SpineJournal.

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

From in vivo testing, overall (mean � SD) axial and sagittaltranslational errors were (1.79 � 1.41 mm) and (2.68 � 2.26 mm)and axial and sagittal angular errors were (3.63� � 2.92�) and(4.65� � 3.36�), respectively (Figure 7). In univariate analyses,there were no statistically significant differences in absolutenavigation errors between cadaveric and clinical studies. MISscrews showed increased quantitative error, relative to matchedopen thoracolumbar controls, in axial translation (1.79 � 1.41mm vs. 0.97 � 0.89 mm; P ¼ 0.004), axial angle (3.63� � 2.92�

vs. 2.73� � 2.09�; P ¼ 0.032), sagittal translation (2.68 � 2.26mm vs. 1.05 � 0.95 mm; P < 0.001), and sagittal angle (4.65� �3.36� vs. 2.82� � 2.29�, P ¼ 0.006). These differences persistedwhen matching was performed by pedicle diameter rather thanby age/gender/spinal level. However, in general linear modeling

WORLD NEUROSURGERY 125: e863-e872, MAY 2019

including distance from the registered level as a covariate, therewere no significant differences in any quantitative errorsbetween MIS and open thoracolumbar cases. All in vivo screwswere placed 0, 1, or 2 vertebral levels from the registered level.Increasing distance between the instrumented and registeredlevels correlated positively with increased axial translationalerror (Pearson correlation coefficient 0.534; P ¼ 0.007).

Radiographic Navigation AccuracyFrom 2 independent raters, an average of 94.5% of in vivo screwswere rated as acceptable on the 2-mm grade, and 100% ratedacceptable by the Heary classification. Three screws were rated aspoor (grade >2) on the 2-mm classification by 1 or both raters. All3 screws were placed in lumbar vertebrae (L3 or L4) intentionally

www.journals.elsevier.com/world-neurosurgery e867

Table 1. Characteristics of Registrations with OpticalTopographic Imaging Through Mini-Open Minimally InvasiveSurgery Exposures, in Cadaveric Testing

Level andExposure (mm)

SuccessfulRegistrations(% of Total)

Number of Registered Points(Mean � Standard

Deviation)

T2 85.7 355 � 152

25 � 25 66.7

30 � 30 100.0

35 � 35 75.0

40 � 40 100.0

T6 82.1 608 � 288

25 � 25 80.0

30 � 30 72.7

35 � 35 85.7

40 � 40 100.0

T10 67.6 440 � 234

25 � 25 33.3

30 � 30 88.9

35 � 35 87.5

40 � 40 62.5

L3 58.5 353 � 173

25 � 25 25.0

30 � 30 33.3

35 � 35 66.7

40 � 40 66.7

Overall 71.8 431 � 235

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

with a more lateral starting point, at the junction of the transverseprocess and superior articular process (Figure 8). This is a well-documented technique allowing docking of the awl against thetransverse process/facet junction for tactile feedback in a percu-taneous procedure, to reduce the profile of the screw heads, and toavoid damaging the superior facet capsule.22

There were no critical radiographic breaches, and no neuro-vascular or other clinical complications from any in vivoinstrumentation.

DISCUSSION

The primary purported benefit of CAN for spinal procedures isimproved instrumentation accuracy and, in theory, minimizationof acute and long-term complications from misplaced screws.CAN has been shown to reduce pedicle screw breach rates from12%e40% with freehand or fluoroscopic guidance to <5% with3D CAN.23-28 Improved instrumentation accuracy is seen across all3D CAN techniques, registering to preoperative or intraoperativeimaging, in each of the cervical, thoracic, lumbar, and sacralregions.29-33

e868 www.SCIENCEDIRECT.com WORLD NE

Workflow disturbances continue to limit the use of CAN amongspinal surgeons, although the technology has been adopted morereadily by specialists in MIS and complex deformity surgery, inwhich bony landmarks may not be readily identifiable. AlthoughOTI has been validated previously as a comparably accurate yetsignificantly faster technique of intraoperative navigation relativeto current CAN systems, its usefulness in mini-open procedures,with limited line of sight to exposed anatomy, has beenunproved.16

Our group has previously quantified absolute navigation accu-racy for the first time in current CAN techniques registered topreoperative or intraoperative imaging, as well as in OTI for openthoracolumbar exposures.16 In our current analysis, although notreaching statistical significance, there was a trend towardincreased translational and angular errors for MIS exposures inclinical testing relative to preclinical cadaveric validation. Thissituation is likely caused part by the placement of in vivo screwspercutaneously over K-wires, but without a navigatedscrewdriver, which may have led to deviation of the final screwplacement from the original navigated tract. Furthermore,in vivo screws were placed percutaneously at up to 2 levelsdistant from the registered level, potentially introducing error asa result of intersegmental mobility. Our group has quantifieddirectly the navigation inaccuracy resulting from intersegmentalmotion, showing that instrumentation in vivo up to 2 levelsfrom the level to which the reference frame is clamped is safe.34

Corroborating this theory, absolute translational and angularerrors for in vivo MIS exposures were slightly greater, in bothaxial and sagittal planes, than those obtained using OTI foropen thoracolumbar exposures, but not significantly oncedistance from the registered level was accounted for in generallinear modeling. The impact of distance from the registeredlevel on navigation accuracy was not assessed in the cadavericsetting, because the full spinal exposure allowed segmentalregistration; however, we have shown in our own study that asubstantial component of intersegmental mobility is related torespiration-induced motion, seen of course only in an in vivosetting. The literature on the impact of nonsegmental registrationon navigation accuracy is heterogeneous, both in outcomes as wellas in the metrics used to quantify navigation accuracy.35-37 Thequantitative translational accuracy of OTI for MIS remains within2e3 mm, comparing favorably to the accuracy of commercial CANsystems. Moreover, the radiographic accuracy of screws placedafter MIS-OTI registrations was 100% by the Heary classification,with no clinical complications. The slightly greater quantitativeinaccuracy in MIS versus open OTI procedures is likely related tothe percutaneous placement of screws rather than to the regis-tration itself, because the lack of visual anatomic feedback and theunavailability of a tracked screwdriver, with an untapped screwtract, allow for increased error in screw placement relative to theintended navigation-guided trajectory.The ability of OTI to perform patient-to-image registration is

contingent on the acquisition of sufficient exposed points that canbe correlated, using an iterative closest-point algorithm, to cor-responding points on a preoperative image set. Because thisprocess is most readily performed on bony anatomy, standardmidline mini-open exposures were chosen for this initialdemonstration of feasibility for MIS approaches. Cadaveric

UROSURGERY, https://doi.org/10.1016/j.wneu.2019.01.201

Figure 5. Standard boxplots showing the number ofsurface points registered by optical topographicimaging stratified by exposure size (A), and byregistered level (B), in cadaveric testing. Boxes

represent the first, median, and third quartiles.Whiskers represent 1.5� the interquartile range.*Represents significance at P < 0.05.

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

registrations were performed in the upper, middle, and lowerthoracic spine as well as lumbar spine, where the bulk of MISprocedures are performed.38,39 Pedicle screws were inserted at alllevels. Concurrent cortical trajectory screws were placed at L3,because cortical screws are commonly placed in MIS midlinedecompression and fusion procedures to achieve greater bonypurchase with minimal muscle dissection and soft-tissueretraction.40,41

We found that the number of surface points acquired andregistered by the navigation system correlated positively with thelikelihood of successful registration. The first quartile of registered

Figure 6. Standard boxplots showing the absolutetranslational (A) and angular (B) navigation errors in theaxial and sagittal planes, stratified by registered level

WORLD NEUROSURGERY 125: e863-e872, MAY 2019

points for successful registrations and the third quartile forunsuccessful registrations converged at approximately 325 points.However, the number of registered points did not correlate withany absolute navigation error. In this iteration of OTI, 325 regis-tered points should therefore be targeted as the minimum forsuccessful registration through an MIS exposure, with more pointsincreasing the likelihood of successful registration but not finalnavigation accuracy.Although the number of surface points registered by OTI was

correlated with the likelihood of successful registration, the size ofthe exposure itself was not an independent predictor of

and screw trajectory, in cadaveric testing. Boxesrepresent the first, median, and third quartiles.Whiskers represent 1.5� the interquartile range.

www.journals.elsevier.com/world-neurosurgery e869

Figure 7. Standard boxplot showing the absolutetranslational (A) and angular (B) navigation errors in theaxial and sagittal planes, in clinical in vivo minimally

invasive surgery testing. Boxes represent the first,median, and third quartiles. Whiskers represent 1.5�the interquartile range.

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

registration success. Therefore, although smaller exposures areconsidered the definition of minimally invasive, it is the quality ofthe exposed anatomy rather than the size itself that most affectsthe likelihood of successful registration with OTI. Regions withmore geometric variability, and hence a greater number of uniquepoints that may be used for patient-to-image registration, aremore likely to be registered successfully even with a smaller skinopening than are regions with geometric homogeneity. Forinstance, to achieve a minimum 70% likelihood of registrationsuccess, a minimum 35 � 35 mm exposure was required in thelumbar spine, whereas 25 � 25 mm and 30 � 30 mm exposures

Figure 8. Axial (A) and sagittal (B)multiplanar-reformatted computed tomographyimaging showing a percutaneously inserted right L4screw, with starting point intentionally at the junction

e870 www.SCIENCEDIRECT.com WORLD NE

were sufficient in the thoracic spine. This situation may be causedin part by geometric symmetry in the medial lumbar hemilaminaeand in part by the increased depth of lumbar surgical cavities,resulting in increased shadowing and fewer captured points for anoptically based acquisition system. The latter represents a knowntechnical challenge with OTI, one that is readily rectifiable withmodified camera and projector alignments. The lack of correlationbetween exposure size and registration success was shown only ina cadaveric setting, whereas our in vivo mini-open exposures wereslightly larger. This is an acknowledged limitation of our study,one necessitated in part by the need for an adequate exposure for

of the transverse process and superior articularprocess, and graded as poor by the 2-mmclassification.

UROSURGERY, https://doi.org/10.1016/j.wneu.2019.01.201

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

sufficient osseoligamentous decompression after instrumentationin vivo.There are multiple limitations to our analysis. The armamen-

tarium of MIS spinal surgeons includes tubular ports andretractors, which were unavailable in the cadaveric study for sys-tematic testing of registration success. Percutaneous placement ofinstrumentation is performed best with a tracked screwdriver toensure no deviation from the navigated screw tract. Future studiesof OTI for MIS applications should include percutaneous place-ment of instrumentation distant from the level of registration andreference-frame fixation, accounting for the additional sources ofnavigation error arising from nonsegmental registration.

CONCLUSIONS

Optical machine vision is a novel navigation technique previouslyvalidated for open posterior exposures. OTI is feasible for

WORLD NEUROSURGERY 125: e863-e872, MAY 2019

mini-open MIS exposures in preclinical and initial clinical testing,with comparable radiographic accuracy to that achieved by OTI inopen exposures. The likelihood of successful registration dependson the number of points acquired and registered by the navigationsystem but not exposure size. With the exception of sagittalangular deviation, absolute navigation accuracy is unaffected bythe size of the MIS exposure, or by the number of registeredpoints. Future work exploring the feasibility of OTI registrationthrough tubular minimal-access approaches is warranted.

ACKNOWLEDGMENTS

This research is supported by the Natural Sciences and Engi-neering Research Council of Canada (NSERC) and the CanadaFoundation for Innovation (CFI). Salary support for D.G. wasprovided in part by a Canadian Institutes of Health Research(CIHR) postdoctoral fellowship (FRN 142931).

REFERENCES

1. Hartl R, Lam KS, Wang J, Korge A, Kandziora F,Audige L. Worldwide survey on the use of navi-gation in spine surgery. World Neurosurg. 2013;79:162-172.

2. Rivkin MA, Yocom SS. Thoracolumbar instru-mentation with CT-guided navigation (O-arm) in270 consecutive patients: accuracy rates and les-sons learned. Neurosurg Focus. 2014;36:E7.

3. Wood MJ, McMillen J. The surgical learningcurve and accuracy of minimally invasive lumbarpedicle screw placement using CT basedcomputer-assisted navigation plus continuouselectromyography monitoringea retrospectivereview of 627 screws in 150 patients. Int J SpineSurg. 2014;8, 27-27.

4. Hecht N, Kamphuis M, Czabanka M, et al.Accuracy and workflow of navigated spinalinstrumentation with the mobile AIRO(�) CTscanner. Eur Spine J. 2016;25:716-723.

5. Choo AD, Regev G, Garfin SR, Kim CW. Sur-geons’ perceptions of spinal navigation: analysisof key factors affecting the lack of adoption ofspinal navigation technology. SAS J. 2008;2:189-194.

6. Ryang Y-M, Villard J, Obermüller T, et al.Learning curve of 3D fluoroscopy image-guidedpedicle screw placement in the thoracolumbarspine. Spine J. 2015;15:467-476.

7. Bandiera S, Ghermandi R, Gasbarrini A, BarbantiBrodano G, Colangeli S, Boriani S. Navigation-assisted surgery for tumors of the spine. Eur SpineJ. 2013;22(suppl 6):S919-924.

8. Sakai Y, Matsuyama Y, Nakamura H, et al.Segmental pedicle screwing for idiopathic scoli-osis using computer-assisted surgery. J SpinalDisord Tech. 2008;21:181-186.

9. Dea N, Fisher CG, Batke J, et al. Economic eval-uation comparing intraoperative cone beam CT-based navigation and conventional fluoroscopyfor the placement of spinal pedicle screws: apatient-level data cost-effectiveness analysis. SpineJ. 2016;16:23-31.

10. Al-Khouja L, Shweikeh F, Pashman R, Johnson JP,Kim TT, Drazin D. Economics of image guidanceand navigation in spine surgery. Surg Neurol Int.2015;6:S323-326.

11. Hu W, Tang J, Wu X, Zhang L, Ke B. Minimallyinvasive versus open transforaminal lumbarfusion: a systematic review of complications. IntOrthop. 2016;40:1883-1890.

12. McAnany S, Overley S, Kim J, Baird E, Qureshi S,Anderson P. Open versus minimally invasive fix-ation techniques for thoracolumbar trauma: ameta-analysis. Glob Spine J. 2015;06:186-194.

13. Goldstein CL, Macwan K, Sundararajan K,Rampersaud YR. Perioperative outcomes andadverse events of minimally invasive versus openposterior lumbar fusion: meta-analysis and sys-tematic review. J Neurosurg Spine. 2016;24:416-427.

14. Phan K, Mobbs RJ. Minimally invasive versus openlaminectomy for lumbar stenosis. Spine (Phila Pa1976). 2016;41:E91-100.

15. Costa F, Tosi G, Attuati L, et al. Radiation expo-sure in spine surgery using an image-guided sys-tem based on intraoperative cone-beam computedtomography: analysis of 107 consecutive cases.J Neurosurg Spine. 2016;25:654-659.

16. Jakubovic R, Guha D, Gupta S, et al. High speed,high density intraoperative 3D optical topograph-ical imaging with efficient registration to MRI andCT for craniospinal surgical navigation. Sci Rep.2018;8:14894.

17. Guha D, Jakubovic R, Gupta S, et al. Spinalintraoperative three-dimensional navigation: cor-relation between clinical and absolute engineeringaccuracy. Spine J. 2017;17:489-498.

18. Mathew JE, Mok K, Goulet B. Pedicle violationand Navigational errors in pedicle screw insertionusing the intraoperative O-arm: a preliminaryreport. Int J Spine Surg. 2013;7:e88-94.

19. Kotani Y, Abumi K, Ito M, et al. Accuracy analysisof pedicle screw placement in posterior scoliosissurgery: comparison between conventional fluo-roscopic and computer-assisted technique. Spine(Phila Pa 1976). 2007;32:1543-1550.

www.journals.els

20. Zhang W, Takigawa T, Wu Y, Sugimoto Y,Tanaka M, Ozaki T. Accuracy of pedicle screwinsertion in posterior scoliosis surgery: a com-parison between intraoperative navigation andpreoperative navigation techniques. Eur Spine J.2017;26:1756-1764.

21. Heary RF, Bono CM, Black M. Thoracic pediclescrews: postoperative computerized tomographyscanning assessment. J Neurosurg. 2004;100:325-331.

22. Wong AP, Smith ZA, Stadler JA, et al. Minimallyinvasive transforaminal lumbar interbody fusion(MI-TLIF): surgical technique, long-term 4-yearprospective outcomes, and complicationscompared with an open TLIF cohort. Neurosurg ClinNorth Am. 2014;25:279-304.

23. Wang Y, Xie J, Yang Z, et al. Computed tomog-raphy assessment of lateral pedicle wall perfora-tion by free-hand subaxial cervical pedicle screwplacement. Arch Orthop Trauma Surg. 2013;133:901-909.

24. Bydon M, Mathios D, Macki M, et al. Accuracy ofC2 pedicle screw placement using the anatomicfreehand technique. Clin Neurol Neurosurg. 2014;125:24-27.

25. Castro WH, Halm H, Jerosch J, Malms J,Steinbeck J, Blasius S. Accuracy of pedicle screwplacement in lumbar vertebrae. Spine (Phila Pa1976). 1996;21:1320-1324.

26. Shin BJ, James AR, Njoku IU, Hartl R, Härtl R.Pedicle screw navigation: a systematic review andmeta-analysis of perforation risk for computer-navigated versus freehand insertion. J NeurosurgSpine. 2012;17:113-122.

27. Nottmeier EW, Seemer W, Young PM. Placementof thoracolumbar pedicle screws using three-dimensional image guidance: experience in alarge patient cohort. J Neurosurg Spine. 2009;10:33-39.

28. Amiot LP, Lang K, Putzier M, Zippel H, Labelle H.Comparative results between conventional andcomputer-assisted pedicle screw installation inthe thoracic, lumbar, and sacral spine. Spine (PhilaPa 1976). 2000;25:606-614.

evier.com/world-neurosurgery e871

ORIGINAL ARTICLE

DAIPAYAN GUHA ET AL. OTI FOR SPINAL INTRAOPERATIVE 3D NAVIGATION IN MINI-OPEN APPROACHES

29. Bourgeois AC, Faulkner AR, Bradley YC, et al.Improved accuracy of minimally invasive trans-pedicular screw placement in the lumbar spinewith 3-dimensional stereotactic image guidance: acomparative meta-analysis. J Spinal Disord Tech.2015;28:324-329.

30. Tian NF, Huang QS, Zhou P, et al. Pedicle screwinsertion accuracy with different assistedmethods: a systematic review and meta-analysis ofcomparative studies. Eur Spine J. 2011;20:846-859.

31. Hecht N, Kamphuis M, Czabanka M, et al. Intra-operative Iso-C C-arm navigation in craniospinalsurgery: the first 60 cases. J Neurosurg Spine. 2010;36:E1.

32. Barsa P, Fr}ohlich R, �Sercl M, Buchvald P,Suchomel P. The intraoperative portable CTscanner-based spinal navigation: a viable optionfor instrumentation in the region of cervico-thoracic junction. Eur Spine J. 2016;25:1643-1650.

33. Mason A, Paulsen R, Babuska JM, et al. The ac-curacy of pedicle screw placement using intra-operative image guidance systems. J NeurosurgSpine. 2014;20:196-203.

34. Guha D, Jakubovic R, Gupta S, et al. Intraoperativeerror propagation in 3-dimensional spinal naviga-tion from nonsegmental registration: a prospectivecadaveric and clinical study. Global Spine J. 2018.https://doi.org/10.1177/2192568218804556.

e872 www.SCIENCEDIRECT.com

35. Shimizu M, Takahashi J, Ikegami S, Kuraishi S,Futatsugi T, Kato H. Are pedicle screw perforationrates influenced by registered or unregisteredvertebrae in multilevel registration using a CT-based navigation system in the setting of scoli-osis? Eur Spine J. 2014;23:2211-2217.

36. Papadopoulos EC, Girardi FP, Sama A,Sandhu HS, Cammisa FP. Accuracy of single-time,multilevel registration in image-guided spinalsurgery. Spine J. 2005;5:263-267 [discussion: 268].

37. Scheufler K-M, Franke J, Eckardt A, Dohmen H.Accuracy of image-guided pedicle screw place-ment using intraoperative computed tomography-based navigation with automated referencing. PartII: thoracolumbar spine. Neurosurgery. 2011;69:1307-1316.

38. Banczerowski P, Czigléczki G, Papp Z, Veres R,Rappaport HZ, Vajda J. Minimally invasive spinesurgery: systematic review. Neurosurg Rev. 2015;38:11-26.

39. Smith ZA, Fessler RG. Paradigm changes in spinesurgeryeevolution of minimally invasive tech-niques. Nat Rev Neurol. 2012;8:443-450.

40. Phan K, Hogan J, Maharaj M, Mobbs RJ. Corticalbone trajectory for lumbar pedicle screw place-ment: a review of published reports. Orthop Surg.2015;7:213-221.

WORLD NEUROSURGERY, http

41. Wray S, Mimran R, Vadapalli S, Shetye SS,McGilvray KC, Puttlitz CM. Pedicle screw place-ment in the lumbar spine: effect of trajectory andscrew design on acute biomechanical purchase.J Neurosurg Spine. 2015;22:503-510.

Conflict of interest statement: V.X.D.Y. is co-founder andChief Scientific Officer of 7D Surgical Inc., a companylicensing the OTI technology described in this article. Thereare no material or financial conflicts of interest arising fromthis study. The remaining authors have no relevant conflictsof interest to disclose. This research is supported by theNatural Sciences and Engineering Research Council ofCanada (NSERC) and the Canada Foundation for Innovation(CFI). Salary support for D.G. was provided in part by aCanadian Institutes of Health Research (CIHR) postdoctoralfellowship (FRN 142931).

Received 15 October 2018; accepted 21 January 2019

Citation: World Neurosurg. (2019) 125:e863-e872.https://doi.org/10.1016/j.wneu.2019.01.201

Journal homepage: www.journals.elsevier.com/world-neurosurgery

Available online: www.sciencedirect.com

1878-8750/$ - see front matter ª 2019 Elsevier Inc. Allrights reserved.

s://doi.org/10.1016/j.wneu.2019.01.201