Intra-operative correction of brain-shift

1 23

�� !�"#$ ��#�%�&��'(��)�*��*��+,��-��!.��!��'��

��

�� !��"�

1 23

Your article is protected by copyright andall rights are held exclusively by Springer-Verlag Wien. This e-offprint is for personaluse only and shall not be self-archivedin electronic repositories. If you wish toself-archive your article, please use theaccepted manuscript version for posting onyour own website. You may further depositthe accepted manuscript version in anyrepository, provided it is only made publiclyavailable 12 months after official publicationor later and provided acknowledgement isgiven to the original source of publicationand a link is inserted to the published articleon Springer's website. The link must beaccompanied by the following text: "The finalpublication is available at link.springer.com”.

CLINICAL ARTICLE - VASCULAR

Intra-operative correction of brain-shift

Ingerid Reinertsen & Frank Lindseth &

Christian Askeland & Daniel Høyer Iversen &

Geirmund Unsgård

Received: 31 August 2013 /Accepted: 22 February 2014 /Published online: 3 April 2014# Springer-Verlag Wien 2014

AbstractBackground Brain-shift is a major source of error inneuronavigation systems based on pre-operative images. Inthis paper, we present intra-operative correction of brain-shiftusing 3D ultrasound.Methods Themethod is based on image registration of vesselsextracted from pre-operative MRA and intra-operative powerDoppler-based ultrasound and is fully integrated in theneuronavigation software.Results We have performed correction of brain-shift in theoperating room during surgery and provided the surgeon with

updated information. Here, we present data from seven clin-ical cases with qualitative and quantitative error measures.Conclusion The registration algorithm is fast enough to pro-vide the surgeon with updated information within minutes andaccounts for large portions of the experienced shift. Correctionof brain-shift can make pre-operative data like fMRI and DTIreliable for a longer period of time and increase the usefulnessof the MR data as a supplement to intra-operative 3D ultra-sound in terms of overview and interpretation.

Keywords Brain-shift . Neuronavigation . Ultrasound .MRI

Introduction

Accurate mapping between images and the patient on theoperating table is the cornerstone of any neuronavigationsystem. The mapping between the pre-operative images andthe patient is generally computed after immobilization of thepatient’s head, using fiducial markers, anatomical landmarks,surface matching techniques or a combination of these. Thisimage-to-patient registration is rarely completely accurate butenables navigation for planning of surgery and placement ofthe craniotomy. However, during the procedure the brain willshift and deform, and the spatial relationship with the pre-operative images will be altered even more. Consequently, thenavigational accuracy is reduced and the surgeon can nolonger fully trust the spatial information given by the naviga-tion system. This is the situation in the commercial systemsthat do not have intra-operative imaging capabilities. Theamount of brain-shift depends on several factors such as lossof cerebro-spinal fluid (CSF), gravity, the size and location ofthe surgical target, the amount of resection and administrationof drugs. A number of groups have evaluated the magnitudeand spatial distribution of brain-shift. One of the first studiesdescribed the deformation occurring between pre-operative

I. Reinertsen (*) : F. Lindseth : C. AskelandSINTEF, Medical Technology, P.B. 4760 Sluppen, 7465 Trondheim,Norwaye-mail: [email protected]

I. Reinertsen :D. H. IversenMI-Lab, Norwegian University of Science and Technology (NTNU),Trondheim, Norway

F. LindsethDepartment of Computer and Information Science, NorwegianUniversity of Science and Technology (NTNU), Trondheim, Norway

D. H. IversenDepartment Circulation andMedical Imaging, Norwegian Universityof Science and Technology (NTNU), Trondheim, Norway

G. UnsgårdDepartment of Neuroscience, Norwegian University of Science andTechnology (NTNU), Trondheim, Norway

G. UnsgårdDepartment of Neurosurgery, St. Olav University Hospital,Trondheim, Norway

I. Reinertsen : F. Lindseth : C. Askeland :G. UnsgårdNational Competence Center for Ultrasound and Image guidedTherapy, St. Olav University Hospital, Trondheim, Norway

Acta Neurochir (2014) 156:1301–1310DOI 10.1007/s00701-014-2052-6

Author's personal copy

imaging and the start of the surgical resection in 21 patients andfound maximum displacements of more than 10 mm in one-half of the patients [1]. In another early study of brain shift, 58landmarks in 14 patients were tracked and resulted in a meandisplacement of 10.9 mm with a maximum of 24.6 mm [2].More recent studies of brain shift during placement of elec-trodes for deep brain stimulation have also been performed [3,4]. In [3] they found maximum displacements of more than10 mm even in a procedure with a relatively small craniotomywithout any resection of tissue. They also found that the dis-placement of the cortical surface was larger than the displace-ment of deep structures. In [4] they found displacements up to4 mm in 28 patients and concluded like [3] that brain shift inthis type of surgery is closely related to the amount of CSF loss.A study of brain-shift in patients with brain tumors was pre-sented in [5]. They evaluated the brain shift in 61 patients withsupratentorial lesions and found a maximum displacement of15.2 mm at the end of surgery. They also found significantcorrelations between brain shift and tumor volume, midlineshift and the size of the craniotomy, respectively.

With potential displacements of more than 1 cm, brain-shiftrepresents a major source of error in neuro-navigation systemsbased on pre-operative images. Due to the heterogeneous natureof cerebral lesions and the non-uniform distribution of brain-shift throughout the brain, pre-operative prediction of theamount and course of brain-shift seems unlikely as pointed outin [5]. Consequently, acquisition of intra-operative images dur-ing surgery is an attractive solution. Intra-operative MRI canprovide the surgeonwith updated information during surgery [6,7]. The newest 3 Tesla intra-operative MR scanners providehigh quality images within a reasonable timeframe. In somecenters, this includes the ability to acquire functional MRI(fMRI) [8] and diffusion tensor imaging (DTI) during surgery[9]. However, this solution still presents a number of limitationsincluding high cost resulting from not only the MR scanneritself, but also the need for a large dedicated operating room. Inaddition, positioning of the patient on the operating table toallow MR imaging is complex and the transfer of the patientto the scanner or vice versa presents challenges for both thesurgeons and the anesthesiologists. High field magnets may alsopose a safety hazard [10, 11]. In this context, intra-operativeultrasound (US) presents several advantages. Ultrasound imag-ing is safe, cheap, portable and provides real time imaging evenfor blood flow. In addition, there is no need for dedicatedoperating rooms. However, the interpretation of anatomy andpathology in ultrasound images still seem to present an obstaclefor more widespread use. This challenge can partly be solved byincorporating ultrasound images into the neuronavigation sys-tem and visualize corresponding views of pre-operative MRdata and intra-operative US data [12]. While intra-operativeultrasound imaging is useful for acquisition of updated informa-tion about anatomy, pathology and the degree of resection, theultrasound data present some limitations. The ultrasound data

cover only a limited part of the brain around the surgical target,and no functional data are available. Even with high qualityintra-operative ultrasound imaging, structural MR images arenecessary to provide anatomical overview and ease the interpre-tation of the ultrasound images. Pre-operative fMRI and DTIdata might be crucial to identify important functional regionsand white matter tracts. Right from the beginning of surgery andthroughout the procedure, inaccurate registration between pre-operative images and the patient and subsequent brain-shift willresult in potentially large errors in the navigation based on pre-operative MR data. Navigation based on recently acquiredultrasound, however, will remain accurate [13, 14] because thedata is not associated with the image-to-patient registration error.Navigation based on recently acquired ultrasound data is asso-ciated with an error of less than 1.5 mm, where ultrasound probecalibration is the main contributing factor.

In order to correct the pre-operative MR data to betterreflect the intra-operative reality, several image registrationtechniques have been proposed. These methods register pre-operative T1-weighted MR images with intra-operative B-mode ultrasound images by aligning hyper-echogenic struc-tures [15], gradients [16] or a combination of intensities andgradients [17]. Validation of these methods has been per-formed using between two and 14 retrospective clinicaldatasets. No intra-operative validations of registrationmethods have been presented. Due to the very different imagecharacteristics of MR and ultrasound data, direct registrationbetween T1 MR and B-mode ultrasound is challenging. Wehave, therefore, previously suggested the use of blood vesselsas features for the registration algorithm [18, 19]. The vasculartree in the brain is relatively easy to extract from both ultra-sound (power Doppler imaging) data and MR angiography(MRA) and are consequently well suited for image registra-tion. This method has previously been validated using simu-lated displacements, phantom data and retrospective clinicaldata. In this paper, we present the full integration of the semi-automatic vessel based registration technique into aneuronavigation system and the results of the application ofthe method in the operating room during surgery. To ourknowledge, this is the first report of an ultrasound-basedregistration method to correct for brain-shift running in theoperating room during surgery. The registration method, theintegration into the neuro-navigation system and intra-operative correction of brain-shift in seven clinical cases arepresented in the following sections.

Methods

Intra-operative setup and neuronavigation support

The intra-operative setup used in this study is shown in Fig. 1.The complete system consists of two racks: an in-house

1302 Acta Neurochir (2014) 156:1301–1310


navigation system called CustusX [20] (SINTEFMedical tech-nology, Trondheim, Norway) and a GEVingmed E9 ultrasoundscanner with a GE 11 L linear array transducer (GE VingmedUltrasound, Horten, Norway). The navigation rack assemblesthe main hardware components including a navigation comput-er with display, an optical tracking system (Polaris spectra,Northern Digital Inc., Waterloo, Canada) and a frame grabber(Epiphan systems Inc., Ottawa, Canada). This simplifies logis-tics in the operating room where space is limited. The infraredcamera and the 27″navigation screen are both mounted onflexible arms in order to improve the intra-operative setup.

The optical tracking system is used to acquire the positionand orientation of navigation localizers with reflective spheresrelative to a reference localizer mounted on the operating table(Fig. 1). Localizers are attached to the navigation pointer, theultrasound probes, the ultrasonic surgical aspirator and thebiopsy forceps enabling visualization of all these instrumentsrelative to both pre-operative and intra-operative images onthe navigation screen. The ultrasound scanner transfers im-ages to the navigation computer via a frame grabber. Theposition and orientation of the individual 2D frames can bedetermined by combining the tracking data from the localizermounted on the ultrasound probe and a pre-calibration of thesame probe [21]. Following a freehand scan that covers theregion of interest, the tracked 2D ultrasound images are auto-matically reconstructed into a 3D volume that becomes avail-able for navigation within a few seconds. The navigationscreen during an ultrasound acquisition is shown in Fig. 2.

In this study we have used the navigation platform CustusXfor evaluating new functionality and conducting research in the

operating room. The system can be used for image-guidedsurgery in general, but the main focus of the application isintra-operative ultrasound-based navigation. CustusX is de-signed to be easy to use in the operating room. The mostimportant features are easily accessible at all times, while moreadvanced features are hidden during normal use. A set of pre-defined configurations is used to quickly switch between thestages of the surgical workflow: Data import and preparation,Registration, Planning, Navigation, and UltrasoundAcquisition. A photo of the CustusX-rack and the ultrasoundscanner in the operating room is shown in Fig. 3.

Patients and data

Illustrative cases To verify the accuracy and usefulness ofintra-operative brain-shift correction in various neurosurgicalprocedures, the proposed vessel-based MR-to-ultrasound reg-istration method was evaluated during the following sevenclinical cases:

1. Aneurysm: Right middle cerebral artery bifurcationaneurysm.

2. AVM: Arteriovenous malformation (AVM) located in theleft frontal lobe.

3. Tumor: High-grade brain tumor in the left occipital/parietal lobe.

4. AVM: AVM located in the left frontal lobe5. Tumor: Low-grade tumor located in the right insula6. Tumor: Low-grade tumor located in the left temporal lobe7. Tumor: Low grade tumor located in the left frontal lobe

The study was approved by the local ethics committee, andthe patients gave informed consent prior to the procedure.

Pre-operative MR imaging Prior to MR imaging, five self-adhesive fiducial markers were glued to the patient’s skin. Pre-operative MR images included in all cases one 3D time-of-flight (TOF) MRA sequence covering the entire head (voxelsize: 0.86×0.86×0.90 mm3), and one TOF MRA sequencewith reduced field of view in the z-direction (voxel size:0.26×0.26×0.50 mm3) in order to have improved resolutionin the region of interest. For the tumor patients, pre-operativeMR imaging also included a Gd-enhanced T1-weighted im-age (voxel size: 0.5×0.5×1 mm3) and a fluid-attenuated in-version recovery (FLAIR) (voxel size: 1.0×1.0×1.2 mm3)image. For the AVM and aneurysm patients, the MRAvolumecovering the entire head was used for fiducial-based image-to-patient registration. For the tumor patient, the Gd-enhancedT1-weighted image was used for image-to-patient registration.

Intra-operative ultrasound imaging Intra-operative powerDoppler-based ultrasound imaging was performed on theintact dura following the opening of the skull. We used an

Fig. 1 Intra-operative setup with ultrasound scanner and CustusX rack.In the operating room the navigation screen is mounted on a flexible arm

Acta Neurochir (2014) 156:1301–1310 1303


11 L linear probe attached to the Vivid E9 ultrasound scanner(GE Vingmed Ultrasound, Horten, Norway) for all acquisi-tions. Prior to image acquisition, the probe was draped and anavigation localizer was attached to the probe handle. In orderto optimize the acoustic contact between the probe and thedura, sterile ultrasound gel was applied to the dural surface.We acquired power Doppler-based ultrasound data in all sevencases. For the tumor patients, we also acquired B-mode data inorder to visualize the extent of the tumor.

Registration method

The registration method is based on the use of bloodvessels segmented from pre-operative MRA and intra-operative power Doppler ultrasound. The segmentationof pre-operative MRA was performed either by simplethresholding directly in the navigation software, or bythe region growing segmentation method implementedin ITK-Snap with default settings [22]. The intra-operative ultrasound data were segmented by simplethresholding. Following segmentation of the vasculartree from both modalities, we extracted the vessel cen-terlines using the algorithm proposed in [23] and imple-mented in ITK [24]. An example of vessel centerlinesbefore and after registration is shown in Figs. 4. Theresulting centerlines from each dataset were then regis-tered using a modified version of the iterative closestpoint algorithm (ICP) [25] and can be summarized infive steps:

1. Finding point correspondences: Every point in the mov-ing volume is matched to the closest point in the fixedvolume.

2. Sorting and rejecting point pairs: The least trimmedsquares (LTS) algorithm [26] is used to reduce the influ-ence of incorrect pairings on the final transformation. Thepoint pairs are sorted based on the distance between themoving and fixed point. The point pairs corresponding tothe largest distances are excluded from the estimation ofthe transformation.

Fig. 2 CustusX during ultrasound acquisition (aneurysm-case). The real-time 2D image from the ultrasound scanner (left), 3D visualization withthemoving ultrasound probe and the pre-operativeMR volume (top right)

and extent of 2D ultrasound image in relation to the pre-operative MRvolume (bottom right)

Fig. 3 Neuronavigation with CustusX during the aneurysm operation:the navigation screen (right), the ultrasound scanner (middle) and screenswith the real time image from the microscope (left). The navigationpointer used by the surgeon to identify the different structures can beseen both on the navigation screen and in the real time images from themicroscope

1304 Acta Neurochir (2014) 156:1301–1310


3. Finding the transformation: A rigid transformation iscomputed based on the remaining point pairs.

4. Transforming points: The resulting transformation is ap-plied to the moving points

5. Error estimation: The previous five steps are iterated untilthe difference in mean distance between successive itera-tions is smaller than 0.001 mm or a maximum number of100 iterations are reached.

CustusX has support for a range of image-to-image regis-tration algorithms. They are all available from the Registrationworkflow stage. First, the input data are converted to center-lines. For clear-cut cases this conversion can be performed in afew clicks, but each step can be fine-tuned if necessary.Second, the actual registration is performed. The position andorientation of the different datasets can be linked in the soft-ware. This is particularly useful when a given dataset (e.g., MRangiography) is used for image registration. The position andorientation of the datasets previously linked to this dataset (e.g.,MR T1 and FLAIR) will then be updated automatically as thetransformation is applied. The registration can be verified usingan undo/redo function with corresponding visualization.

Evaluation of the registration accuracy

The registration accuracy was first evaluated intra-operativelyby qualitative assessment of the alignment of the vesselsextracted from MRA and power Doppler-based ultrasoundin the navigation software. The surgeon then used the trackednavigation pointer and identified landmarks on the vessels.The correspondence between the ultrasound dataset, thecorrected MRA dataset and the known location of the pointerin the physical space were then used by the surgeon to eval-uate the registration result and decide if the corrected MRAimage and other pre-operative MR images linked to theMRA,were accurate enough for navigation. Finally, we identifiedfive landmarks in each dataset (MRA and ultrasound) post-

operatively and computed the distances between landmarksbefore and after registration in order to quantitatively assessthe registration accuracy.

Results

Qualitative results In all seven cases, the surgeon judged thepre-operative MR images to be too inaccurate for navigationbefore acquisition of ultrasound data. Thus, the MR imagescould not be trusted in the identification of the feeding arteriesand draining veins of the AVMs, in the localization of theaneurysm and last, but not least, in the identification of thetumor boundary for the tumor cases. Following ultrasoundimaging and correction of brain-shift, the visual alignment ofpre-operative MR data and intra-operative ultrasound datawas clearly improved in all cases. Screenshots from the nav-igation screen before and after correction are shown in Figs. 5and 6. After correction, the surgeon judged the pre-operativedata to be accurate enough for navigation in all seven cases.

Quantitative results Post-operatively, we identified five land-marks in both datasets for all seven patients. An example oflandmarks placed in corresponding ultrasound and MR vol-umes is shown in Fig. 7. The distances between correspondinglandmarks before and after registration are shown in Table 1.As can be seen in the table, the mean distance has beenreduced from 7.39 mm to 2.77 mm. The distances beforeand after registration were compared using a pairedStudent’s t-test. The difference is significant with p<0.001.

Discussion

In this paper, we have shown that correction of brain shiftduring neurosurgical procedures can be successfully

Fig. 4 An example of vesselcenterlines used in the registrationalgorithm (AVM-case): MRA inred and ultrasound in green

Acta Neurochir (2014) 156:1301–1310 1305


performed by MR-to-ultrasound registration. To our knowl-edge, this is the first report of a semi-automatic ultrasound-based registration method to correct for brain-shift running inthe operating room during surgery, and an example of a

successful transfer of an image registration technique to clin-ical practice.

A large number of image processing algorithms are devel-oped and validated using phantom or retrospective clinical

Fig. 5 Snapshots from the navigation system during surgery before (toprow) and after (bottom row) correction of the MR data. The surgeon ispointing to the tumor border as can be seen from the ultrasound slices

(right column) and both MR FLAIR (middle column) and MR angiogra-phy (left column, vessels close to the tumor border) seems to be consid-erable more aligned after the shift correcting registration

1306 Acta Neurochir (2014) 156:1301–1310


data only, and never transferred to the clinic. Successfulclinical validation and application in “real life” cases requiresnot only a fast and accurate algorithm, but also careful inte-gration into the navigation system and the clinical workflow.Issues such as graphical user interface (GUI), adapted visual-ization and computation time are crucial factors. The registra-tion method presented here and the navigation systemCustusX make it possible for the surgeon to obtain correctedMR data within minutes after the acquisition of the ultrasoundimages. While several steps in the registration pipeline arefully automatic, the method still requires some limited userinteraction. This includes segmentation of the vascular treefrom pre-operative MRA, segmentation of ultrasound data,centerline extraction and the actual registration. As explainedpreviously, the design of the navigation software has greatlysimplified these steps, and they can now be performed withina few minutes using the same toolbar window in CustusX.The tuning of processing parameters has been reduced to astrict minimum, and default values based on our experiencewith a large number of datasets have been defined. Thedevelopment of the user interface through a number of labexperiments using retrospective data and extensive experiencewith the use of ultrasound and neuronavigation systems in theoperating room have been major factors in the successful useof the registration method during surgery. In addition, The

CustusX system facilitates qualitative comparison of imagedatasets by combined 2D and 3D visualization. To date, thebiggest uncertainty for successful registrations is the quality ofthe input data. This applies to theMRA as well as to the powerDoppler ultrasound. As the registration algorithm depends onfinding corresponding vessels in the two datasets to bealigned, noise, missing data and vessels that are smearedtogether represent major challenges and potential sources oferror. Improved quality and better definition of the vessels inultrasound imaging is an active field of research in our as wellas other groups. With higher and more consistent imagequality, the registration method can be made fully automatic.Other possible improvements that are under investigation are3D reconstruction algorithms especially adapted to ultrasoundpower Doppler data as well as more advanced segmentationand centerline extraction for MRA and ultrasound data. Therequirement still being that the algorithms can run withinseconds in the operating room.

Quantitative measurements of registration accuracy duringsurgery are challenging. The qualitative evaluation by com-bined 2D and 3D visualization of the datasets in question isstraightforward in our system (see Figs. 5 and 6). In addition,the surgeons get an impression of the accuracy by identifyingknown structures with the navigation pointer and comparingwith the datasets on the navigation screen. For this paper, we

Fig. 6 MRA in red and ultrasound in grey before (top row) and after (bottom row) registration: Aneurysm patient (left), AVMpatient (middle) and tumorpatient (right)

Acta Neurochir (2014) 156:1301–1310 1307


have also performed quantitative measurements post-operatively. Our error measure is based on the Euclidiandistance between landmarks identified on the vessels in theMRA and ultrasound volumes. As the registration algorithmuses the vascular structures to compute the transformation,this error measure is somewhat biased. However, in AVM andaneurysm surgery the vascular structures are of outmost im-portance and the correct position and orientation of the vesselsare, therefore, essential. In tumor surgery, the spatial relation-ship between the tumor and the vascular tree can also becritical. In tumor surgery, however, the alignment of the tumorboundary in MR and ultrasound is also an important errormeasure. The distance between vessels before and after regis-tration show that we are able to reduce the distance betweenthe two datasets considerably in all cases. The distances beforeand after registration were compared using a paired t-test andthe result is statistically significant with p<0.001. Some casespresent large amounts of non-linear deformation that will notbe captured by the rigid registration. The future integration ofnon-linear registration will probably improve the results insuch cases.

The experienced misalignment between pre-operative MRimages and intra- operative ultrasound data during surgery isdue partly to errors related to the point based image-to-patientregistration and partly to brain-shift. The proposed methodcorrects for rigid misalignment caused by both sources. Brain-shift is in general highly non-linear and cannot be completely

accounted for by a simple rigid registration. However, ourprevious study [19] shows that rigid registration in some casescan account for a large part of the experienced mismatch. Incases with inaccurate patient registration and possibly in caseswith significant shift in the direction of gravity due to CSFleakage, the rigid component of the transformation will beimportant. In the near future, non-linear vessel based registra-tion will be implemented in the navigation system and vali-dated intra-operatively. This will probably further improve theregistration accuracy in a number of cases such as the AVMand tumor cases presented here. We will also investigate thecorrection of fMRI and DTI data. As the different datasets canbe linked in the software, the registration result obtained usingthe MRA data can be automatically applied to datasets likefMRI and DTI. As these datasets contain information that isnot otherwise easily available to the surgeon, this will proba-bly be an important application of the registration method inthe future. Correction of fMRI and DTI will make thesedatasets considerably more accurate and, thus, usable for alonger period of time during surgery. This is particularlyimportant for surgery in eloquent areas.

In this paper, we have presented intra-operative registrationbetween pre-operative MR and the first ultrasound data ac-quired on the intact dura. Ultrasound data are usually acquiredat least three times during surgery, sometimes as many as tentimes, whenmultiple resection controls are needed. This paperdescribes, thus, the first step in a series of possible corrections

Fig. 7 Example of landmarksplaced on the vessels inultrasound (left) and MR (right)

Table 1 Mean distances±std be-tween five landmarks in the MRAand ultrasound volumes beforeand after brain-shift correction.Maximum distances are given inparentheses

Patient # Lesion type Mean dist. before corr. (mm) Mean dist. after corr. (mm)

1 AVM 9.07±1.77 (10.85) 2.87±0.93 (3.91)

2 Aneurysm 7.33±1.24 (9.04) 1.42±0.86 (2.43)

3 Tumor 5.55±2.01 (8.04) 1.80±1.35 (3.44)

4 AVM 4.39±2.02 (7.12) 3.04±0.73 (4.03)

5 Tumor 7.62±0.71 (8.32) 4.67±0.53 (5.38)

6 Tumor 10.06±1.96 (12.33) 2.03±0.90 (3.33)

7 Tumor 7.71±1.37 (9.20) 3.61±1.44 (5.59)

Total 7.39±1.58 (12.33) 2.77±0.96 (5.59)

1308 Acta Neurochir (2014) 156:1301–1310


for brain-shift during surgery. In the following steps bothMR-to-ultrasound (multimodal) and ultrasound-to-ultrasound(monomodal) registrations are possible. These possibilitieswill be further explored in the near future and evaluated interms of registration accuracy, robustness and clinical useful-ness. We will also explore the use of ultrasound B-modeimages alone or in combination with power Doppler data forcorrection of brain-shift. This might be a valuable addition incases where the surgical target is located in regions with fewor no vessels suitable for vessel based registration. The goal offuture developments will still be to capture brain-shift using3D ultrasound and provide accurate navigation during theentire procedure in a broad range of neurosurgical operations.

Conclusions

We have presented the successful application of brain-shiftcorrection in the operating theater using image registrationbetween pre-operativeMR data and intra-operative ultrasounddata. The method is fully integrated into the neuronavigationsystem and can be used for any surgical case where 3D MRAis available pre-operatively and power Doppler imaging canbe performed during the procedure. This system has thepotential to increase the usability of pre-operative MR, whichis particularly important for data that currently cannot beacquired using intra-operative ultrasound (e.g., fMRI andDTI). Accurate correspondence between MR and ultrasounddata during most of the surgical procedure will also ease theinterpretation of the ultrasound images and the MR data willprovide a more accurate and realistic anatomical overview ofthe surgical field not available using ultrasound data alone.More intuitive and easier interpretation of intra-operative ul-trasound images will hopefully also make this image modalitymore attractive and result in more widespread use inneurosurgery.

Disclosure This work received financial and material support fromMI-Lab, Center for research-based innovation, NTNU, Trondheim, Norway,National Competence Center for Ultrasound and Image guided Therapy,St. Olav University Hospital, Trondheim, Norway and the Department ofCirculation and Medical Imaging, NTNU, Trondheim, Norway.

Conflict of interest None.

References

1. Hill DLG, Maurer CR, Maciunas RJ, Barwise JA, Fitzpatrick JM,Wang MY (1998) Measurement of Intraoperative Brain SurfaceDeformation under a Craniotomy. Neurosurgery 43:514–528

2. Roberts DW, Hartov A, Kennedy FE, Miga MI, Paulsen KD (1998)Intraoperative Brain Shift and Deformation: A Quantitative Analysisof Cortical Displacement in 28 Cases. Neurosurgery 43:749–760

3. Elias WJ, Fu K-M, Frysinger RC (2007) Cortical and subcorticalbrain shift during stereotactic procedures. J Neurosurg 107:983–988

4. Miyagi Y, Shima F, Sasaki T (2007) Brain shift: an error factor duringimplantation of deep brain stimulation electrodes. J Neurosurg 107:989–997

5. ReingesMHT, Nguyen HH, Krings T, Hütter BO, Rohde V, GilsbachJM (2004) Course of brain shift during microsurgical resection ofsupratentorial cerebral lesions: limits of conventionalneuronavigation. Acta Neurochir 146:369–377

6. Jolesz FA (2011) Intraoperative imaging in neurosurgery: where willthe future take us? Acta neurochirurgica. Supplement 109:21–25

7. Nimsky C, Ganslandt O, von Keller B, Romstöck J, Fahlbusch R(2004) Intraoperative High-Field-Strength MR Imaging:Implementation and Experience in 200 Patients1. Radiology 233:67–78

8. Feigl GC, Safavi-Abbasi S, Gharabaghi A, Gonzalez-Felipe V, ElShawarby A, Freund HJ, Samii M (2008) Real-time 3T fMRI data ofbrain tumour patients for intra-operative localization of primarymotor areas. Eur J Surg Oncol J Eur Soc Surg Oncol Br AssocSurg Oncol 34:708–715

9. Vassal F, Schneider F, Nuti C (2013) Intraoperative use of diffusiontensor imaging-based tractography for resection of gliomas locatednear the pyramidal tract: comparison with subcortical stimulationmapping and contribution to surgical outcomes. Br J Neurosurgdoi:10.3109/02688697.2013.771730

10. Foroglou N, Zamani A, Black P (2009) Intra-operative MRI (iop-MR) for brain tumour surgery. Br J Neurosurg 23:14–22

11. Jankovski A, Francotte F, Vaz G, Fomekong E, Duprez T, Van BovenM, Docquier MA, Hermoye L, Cosnard G, Raftopoulos C (2008)Intraoperative magnetic resonance imaging at 3-T using a dual inde-pendent operating room-magnetic resonance imaging suite: develop-ment, feasibility, safety, and preliminary experience. Neurosurgery63:412–424, discussion 424–416

12. Unsgaard G, RyghOM, Selbekk T,Muller TB, Kolstad F, Lindseth F,Hernes TA (2006) Intra-operative 3D ultrasound in neurosurgery.Acta Neurochir 148:235–253, discussion 253

13. Lindseth F, Bang J, Lango T (2003) A robust and automatic methodfor evaluating accuracy in 3-D ultrasound-based navigation.Ultrasound Med Biol 29:1439–1452

14. Lindseth F, Lango T, Bang J, Nagelhus Hernes TA (2002) Accuracyevaluation of a 3D ultrasound-based neuronavigation system.Comput Aided Surg 7:197–222

15. Coupe P, Hellier P, Morandi X, Barillot C (2012) 3D RigidRegistration of Intraoperative Ultrasound and Preoperative MRBrain Images Based on Hyperechogenic Structures. Int J BiomedImaging 2012:1–14

16. Nigris D, Collins DL, Arbel T (2012) Fast and Robust RegistrationBased on Gradient Orientations: Case Study Matching Intra-operative Ultrasound to Pre-operative MRI in Neurosurgery. In:Abolmaesumi P, Joskowicz L, Navab N, Jannin P (eds) InformationProcessing in Computer-Assisted Interventions. (Lecture Notes inComputer Science), vol 7330. Springer Berlin Heidelberg, BerlinHeidelberg, pp 125–134

17. Roche A, Pennec X, Malandin G, Ayache N (2001) RigidRegistration of 3D Ultrasound with MR Images: A New ApproachCombining intensity and Gradient Information. IEEE Trans MedImaging 20:1038–1049

18. Reinertsen I, DescoteauxM, Siddiqi K, Collins DL (2007) Validationof vessel-based registration for correction of brain shift. Med ImageAnal 11:374–388

19. Reinertsen I, Lindseth F, Unsgaard G, Collins DL (2007) Clinicalvalidation of vessel-based registration for correction of brain-shift.Med Image Anal 11:673–684

20. Askeland C, Lindseth F, Bakeng JBL, Solberg OV, Tangen GA(2011) CustusX: A Research Application for Image-GuidedTherapy. Insight J 1–8

Acta Neurochir (2014) 156:1301–1310 1309


http://dx.doi.org/10.3109/02688697.2013.771730

21. Lindseth F, Tangen GA, Lango T, Bang J (2003) Probe calibration forfreehand 3-D ultrasound. Ultrasound Med Biol 29:1607–1623

22. Yushkevich PA, Piven J, Hazlett HC, Smith RG, Ho S, Gee JC, GerigG (2006) User-guided 3D active contour segmentation of anatomicalstructures: significantly improved efficiency and reliability.NeuroImage 31:1116–1128

23. Lee BK, Gobbi DG, Peters T (2000) Vascular tree extraction frommra and power doppler us image volumes. Engineering in Medicine

and Biology Society, 2000. Proceedings of the 22nd AnnualInternational Conference of the IEEE, vol 3. IEEE, pp 1731–1733

24. Homann H (2007) Implementation of a 3D thinning algorithm.Insight J 1–4

25. Besl PJ, McKay ND (1992) A method for registration of 3-D shapes.Pattern Anal Mach Intell IEEE Trans 14:239–256

26. Rousseeuw PJ, Leroy AM (2005) Robust regression and outlierdetection. Wiley, Hoboken

1310 Acta Neurochir (2014) 156:1301–1310


Intra-operative correction of brain-shift

Documents

Transcript of Intra-operative correction of brain-shift