Evaluating an Image-Guided Operating Room with Cone Beam CT for Skull Base Surgery

Abstract

Importance  Skull base surgery requires precise preoperative assessment and intraoperative management of the patient. Surgical navigation is routinely used for complex skull base cases; however, the image guidance is commonly based on preoperative scans alone.

Objective  The primary objective of this study was to assess the image quality of intraoperative cone-beam computed tomography (CBCT) within anatomical landmarks used in sinus and skull base surgery. The secondary objective was to assess the registration error of a surgical navigation system based on intraoperative CBCT.

Design  Present study is a retrospective case series of image quality after intraoperative cone beam CT.

Setting  The study was conducted at Toronto General Hospital and Princess Margaret Cancer Centre, University Health Network, Toronto.

Participants  A total of 46 intraoperative scans (34 patients, 21 skull base, 13 head and neck) were studied.

Main Outcome and Measures  Thirty anatomical landmarks (vascular, soft tissue, and bony) within the sinuses and anterior skull base were evaluated for general image quality characteristics: (1) bony detail visualization; (2) soft-tissue visualization; (3) vascular visualization; and (4) freedom from artifacts (e.g., metal). Levels of intravenous (IV) contrast enhancement were quantified in Hounsfield's units (HU). Standard paired-point registration between imaging and tracker coordinates was performed using 6 to 8 skin fiducial markers and the corresponding fiducial registration error (FRE) was measured.

Results  Median score for bony detail on CBCT was 5, remaining at 5 after administration of IV contrast. Median soft-tissue score was 2 for both pre- and postcontrast. Median vascular score was 1 precontrast and 3 postcontrast. Median score for artifacts on CBCT were 2 for both pre-and postcontrast, and metal objects were noted to be the most significant source of artifact. Intraoperative CBCT allowed preresection images and immediate postresection images to be available to the skull base surgeon. There was a significant improvement in mean (standard deviation [SD]) CT intensity in the left carotid artery postcontrast 334 HU (67 HU) ( p  < 10 ⁻¹⁰ ). The mean FRE was 1.8 mm (0.45 mm).

Conclusion  Intraoperative CBCT in complex skull base procedures provides high-resolution bony detail allowing immediate assessment of complex resections. The use of IV contrast with CBCT improves the visualization of vasculature. Image-guidance based on CBCT yields registration errors consistent with standard techniques.

Introduction

Technological advances in endoscopic sinus surgery have allowed increasingly complex pathology of the skull base to be treated in this manner. With the presence of critical neurovascular structures, skull base surgery requires precise preoperative assessment and intraoperative management of the patient. Preoperative imaging is essential to determine the patient's anatomical landmarks and assess the extent of the resection required. Surgical navigation is routinely used for complex skull bases cases and has been shown to reduce complications and improve surgical outcomes. ^{1 2} Despite the advances in surgical navigation, the images used are typically preoperative scans alone, and are unable to capture intraoperative changes due to anatomical deformation and tissue excision.

With the advent of cone-beam computed tomography (CBCT) technology, ^{3 4} intraoperative three-dimensional (3D) imaging is now emerging in many surgical subspecialties. Cone-beam systems using a flat-panel detector provide volumetric imaging without translation of the table or the patient. These imaging devices have been incorporated into hybrid operating rooms across a range of applications, including neurosurgery, ⁵ spine, ⁶ thoracic, ⁷ orthopaedics, ⁸ and head and neck. ⁹ In sinus and skull base surgery, previous studies have demonstrated the benefits of intraoperative CBCT imaging to improve resection completeness ^{1 10} but no study has systematically assessed image quality and navigation accuracy in a clinical setting. The primary objective of this study was to assess the image quality of sinus and skull base anatomical landmarks in intraoperative cone-beam CT (CBCT) scans. The secondary objective was to assess the registration error of a surgical navigation system based on intraoperative CBCT.

Methods

Intraoperative CT Imaging

CBCT imaging was performed with a robotic C-arm system (Artis Zeego, Siemens Healthcare, Erlangen, Germany). As shown in Fig. 1A , the X-ray tube and 40 × 30 cm ² flat-panel detector are mounted on a multiaxis robotic stand that enables a variety of scanning orbits and parked positions. This scanner is located in a multiuse hybrid OR (Guided Therapeutics Operating Room, Toronto General Hospital, University Health Network) which is being used for multiple subspecialties in surgical oncology including otolaryngology, thoracic, and orthopaedics. In this study, the nominal CBCT acquisition consisted of 248 X-ray projections obtained during a 10 seconds, 200-degree rotational orbit, with the X-ray tube passing under the carbon-fiber table. The corresponding radiation dose was approximately 1 mSv, consistent with prior studies. ¹¹ Nominal 3D reconstructions encompassed 25 × 25 × 18 cm ³ using isotropic 0.5 mm ³ voxels.

Fig. 1.

( A ) Intraoperative cone-beam CT imaging (DynaCT) using the Siemens Artis Zeego with carbon-fiber Mayfield head frame (Integra) mounted on carbon-fiber table. ( B ) Coronal and ( C ) sagittal CBCT slices illustrate metal streak artifacts from steel pins which lie superior to surgical area of interest, and analysis of CT (HU) with IV contrast performed using virtual 3D spheres in the carotid artery. ( D ) Endoscopic sinonasal tumor surgery with in-house surgical navigation system registered to intraoperative imaging. 3D, three-dimensional; CT, computed tomography; HU, Hounsfield's units; IV, intravenous.

Cases with neurosurgical access used a radiolucent Mayfield skull clamp (Infinity XR2, Integra, Plainsboro, New Jersey, United States), see Fig. 1A . Positioning of the support arms underneath the skull clamp in this manner allowed for full rotation of the Zeego (Artis Zeego, Siemens Healthcare) during CBCT imaging. We elected to use stainless steel pins, rather than more expensive carbon fiber pins, as the metal artifacts from steel lie superior to anatomical structures in the sinuses and skull base, see Fig. 1B, C .

Intravenous (IV) contrast was administered at most once per case, in patients with no contraindications for iodine-based contrast. The nominal protocol of 110 mL over 160 seconds followed diagnostic imaging protocols for head and neck patients at our institution and was comprised of the following split-bolus sequence: (1) 50 mL at 1.5 cc/sec; (2) a 90-second delay; and (3) 60 mL at 1.5 cc/sec. Injections were typically performed in an antecubital vein using a portable angiographic system (Mark 7 Arterion, Medrad, Warrendale, Pennsylvania, United States). As shown in Fig. 1B, C , levels of contrast enhancement were quantified using a 5-mm diameter spherical region of interest (ROI) in the left carotid artery (adjacent to the sphenoid sinus) using ITK-SNAP, an open-source 3D segmentation package (University of Pennsylvania). ¹² p -Values were calculated using Wilcoxon's rank-sum test.

Surgical Navigation

Image guidance was performed using a custom navigation system comprised of a stereoscopic infrared tracker (Polaris Spectra, NDI, Waterloo, Ontario, Canada) and in-house C + + visualization software (“GTx-Eyes”) ¹³ based on the open-source image-guided surgery toolkit (IGSTK). ¹⁴ Fig. 1C shows the intraoperative navigation setup during endoscopic skull base surgery. The infrared optical tracker was positioned above an endoscopy monitor to provide direct line of sight to optical markers attached to a navigation pointer and patient reference. The robotic Zeego (Artis Zeego, Siemens Healthcare) scanner is shown in a parked position that minimized interference with the standard surgical setup and workflow. The 56-inch bedside monitor presented a “surgical dashboard” from the in-house software platform for integrated visualization of intraoperative imaging, tracked surgical tools, and endoscopic video.

A patient navigation reference was attached to a head strap (Medtronic Navigation, Louisville, Colorado, United States) or, if in use, to the Mayfield. A set of 6 to 8 skin fiducial markers (Medtronic) were placed according to the manufacturer's guidelines on firm parts of the patient's head prior to scanning. Standard paired-point registration between imaging and tracker coordinates was performed by first manually localizing fiducial divots in the image and then touching divots with a navigation pointer, and the corresponding fiducial registration error (FRE) was measured. ¹⁵

Clinical Workflow

The research protocol permitted collection of up to six intraoperative scans over the course of a single operation; however, in the majority of cases 2 to 3 scans were obtained at key clinical milestones. Intraoperative imaging and surgical navigation was implemented to ensure patient and staff safety. First, the surgical field was covered with a transparent sterile drape during imaging, and the patient draping was trimmed and taped below table level to allow unobstructed rotation of the C-arm. A clearance check was then performed prior to imaging, to ensure that the C-arm rotation did not interfere with the standard surgical setup (e.g., anesthesia lines). Finally, surgical tools and anesthesia lines containing artifact-inducing metal were positioned outside of the imaging field when feasible. During imaging, the patient was monitored by the X-ray technician and anesthetist who were both protected by radiation shielding. In accordance with “as low as reasonably achievable” (ALARA) principles of radiation protection, all other members of the operating team left the room. The reconstructed 3D image was available immediately (<30 seconds) for review by surgical staff when they returned to the room.

Patient Population

All patients were recruited under informed consent with institutional ethics approval. Thirty-four patients were included in this study. Of these, 21 were endoscopic procedures in the sinuses and anterior skull base (nine involving neurosurgery), and the remaining 13 were open procedures in head and neck anatomical sites including the mandible, maxilla, and temporal bone who had intraoperative CBCT performed as part of their procedure.

Image Quality Evaluation

Forty-six intraoperative CBCT scans were used for image quality evaluation. Images were reviewed independently by two head and neck radiologists (one senior staff, one fellow) using a diagnostic workstation. The reviewers were free to manipulate viewing settings (e.g., window/level, pan, and zoom). After reviewing each image, a questionnaire was completed to evaluate the visibility of thirty anatomical landmarks and structures within the sinuses and anterior skull base. The image quality of each landmark was assessed using a five-point Likert's scale as follows: 1 = unacceptable; 2 = poor; 3 = satisfactory; 4 = good; and 5 = excellent. Using the same rating scale, the observers also assessed the following four general image quality characteristics: (1) bony detail visualization; (2) soft-tissue visualization; (3) vascular visualization; and (4) freedom from artifacts (e.g., metal).

Results

Patient Demographics

Table 1 demonstrates the patient range, with 33 patients in total, 21 endoscopic procedures, and 12 open procedures. Open procedures were included along with endoscopic procedures to enhance the variation in scans that were being assessed. There were 21 males and 12 females. The mean age was 59.2 years (range: 28–81 years). IV contrast and the surgical navigation system were not used during the open procedures. For the 21 endoscopic procedures, IV contrast was used in 15 cases, and surgical navigation was used in 20 cases (one case was deemed inoperable after endoscopy and imaging, prior to navigation).

Table 1. Study cohort characteristics.

	Clinical			Analysis
	Diagnosis	Procedure	Approach	Image quality	IV contrast	Nav. error
1	Mucosal melanoma	Parotidectomy, maxillectomy, ethmoidectomy	Open	Yes
2	Mandible ORN	Rt. Mandibulectomy	Open	Yes
3	Parotid ca	Parotidectomy, mandibulectomy, temp. bone	Open	Yes
4	Ext. auditory canal SCC	Parotidectomy, temp. bone	Open	Yes
5	High-grade pleomorphic sarcoma	Mandibulectomy, maxillectomy, temp. bone	Open	Yes
6	Maxilla ACC	Total maxillectomy	Open	Yes
7	Maxillary sinus ca	Rt. Maxillectomy	Open	Yes
8	Mandible ca	Bilateral mandibulectomy	Open	Yes
9	Preauricular skin SCC	Partial mandibulectomy, parotid, temp. bone	Open	Yes
10	Biopsy	Transpterygoid to infratemporal fossa	Endo		Yes	Yes
11	Maxillary sinus ca	Lt. maxillectomy	Open	Yes
12	Skull base mass	Transclival/transpterygoid biopsy/resection	Endo	Yes	Yes	Yes
13	Skin/parotid SCC	Parotidectomy, mandibulectomy, temp. bone	Open	Yes
14	Sinonasal mucosal melanoma	Nonoperable: tumor invasion	Endo	Yes	Yes
15	Sinonasal schwannoma	Sinonasal mass	Endo	Yes		Yes
16	CSF leak	Ant. skull base CSF leak	Endo	Yes		Yes
17	Maxilla ACC	Rt. maxillectomy	Open	Yes
18	Sphenoid meningiocele	Repair and drainage	Endo	Yes	Yes	Yes
19	Sphenoid mass	Sphenoidotomy	Endo	Yes		Yes
20	Inverted papilloma	Rt. sinonasal mass resection	Endo	Yes	Yes	Yes
21	Sinonasal carcinoma	Ant. craniofacial resection	Endo		Yes	Yes
22	Sinonasal oligometastatic	Ant. craniofacial resection nasopharyngectomy	Endo	Yes	Yes	Yes
23	SCC ethmoid sinuses	Ant. craniofacial resection	Endo	Yes	Yes	Yes
24	Nasopharyngeal ca	Biopsy in sphenoid, pterygopalatine mass	Endo	Yes	Yes	Yes
25	Mucosal melanoma	Nasal/skull base mass	Endo	Yes	Yes	Yes
26	Nasal capillary hemangioma	Medial maxillectomy	Endo	Yes		Yes
27	Ameloblastoma	Craniofacial resection	Endo	Yes	Yes	Yes
28	Skull base encephalocele	Ant. cranial base recon	Endo			Yes
29	Mucosal melanoma	Medial maxillectomy	Endo		Yes	Yes
30	Skull base mass	Sphenoclival/skull base	Endo		Yes	Yes
31	Sinonasal SCC	Ant. craniofacial resection	Endo			Yes
32	Mixed adeno neuroendocrine ca	Lt. craniofacial resection	Endo		Yes	Yes
33	Nasopharynx adenocarcinoma	Rt. nasopharyngectomy	Endo		Yes	Yes

Abbreviations: ACC, adenoid cystic carcinoma; Ant. anterior; CSF, cerebrospinal fluid; Endo, endoscope; Ext., external; IV, intravenous; Lt., left; Nav, navigation; ORN, osteoradionecrosis; Rt., right; SCC, squamous cell carcinoma.

Image Quality

There were 20 bony landmarks assessed on the CBCT images, see Table 2 . All bony landmarks scored highly on CBCT image assessment pre-and postcontrast. The median overall score for bony detail was 5, both with and without contrast. Fig. 2A, B shows representative images with bony landmarks. Three soft tissue structures (V2, V3, and optic chiasm) were assessed on CBCT, and all had scores of 1 both with and without contrast. The mean overall score for soft tissue on the CBCT was 2 precontrast and 2 postcontrast. Fig. 2C, D shows representative images with soft-tissue landmarks. Seven vascular landmarks were assessed on the CBCT; anterior cerebral artery, carotid genu, carotid artery (petrous portion), basilar artery, jugular bulb, internal maxillary artery and the middle meningeal artery. All vascular landmarks other than the middle meningeal artery were rated as 1 precontrast and 3 postcontrast. The middle meningeal artery scored 1 pre-and postcontrast. The overall assessment of vascular detail was rated as 1 precontrast and 3 postcontrast. Fig. 2E, F shows representative images with vascular landmarks. Median score for artifacts on CBCT were 2 both pre-and postcontrast, and metal objects were noted to be the most significant source of artifact.

Table 2. Anatomical landmarks with and without contrast.

Anatomical landmark	No contrast	With contrast
Clivus	5	5
Petrous apex	5	5
Occipital condyle	5	5
Foramen magnum	5	5
Dens	5	5
Pterygopalatine fossa	5	5
Medial pterygoid plate	5	5
Lateral pterygoid plate	5	5
Foramen ovale	5	5
Foramen spinosum	5	5
Hypoglossal canal	5	5
Frontal sinus	5	5
Crista gali	5	5
Fovea ethmoidalis/cribriform plate	4	4
Anterior clinoid/optic canal	5	5
Anterior ethmoid (bony nipple)	4	4
Foramen rotundum	5	4
Vidian canal	4	4
Foramen lacerum	4	4
Sella turcica	5	5
Anterior cerebral artery	1	3
Carotid genu	1	3
Carotid artery—petrous portion	1	3
Basilar artery	1	3
Jugular bulb	1	3
Internal maxillary artery	1	3
Middle meningial artery	1	1
V2 nerve	1	1
V3 nerve	1	1
Optic chiasm	1	1

Abbreviations: 1, unacceptable; 2, poor; 3, satisfactory; 4, good; 5, excellent.

Fig. 2.

Representative CBCT images showing samples of ( A , B ) bony landmarks; ( C , D ) soft-tissue landmarks; ( E , F ) vascular landmarks. CBCT, cone-beam computed tomography; PPF, pterygopalatine fossa.

Use of IV Contrast

IV contrast was administered in 15 of the 21 endoscopic transnasal procedures to highlight complex vascular anatomy. IV contrast was not routinely given to patients undergoing open procedures. Quantitative analysis of CT (HU) with performed for the endocopic patients that received IV contrast. This was performed using 3D spherical regions of interest (ROI) placed in the carotid artery adjacent to the sphenoid sinus. There was a significant improvement in ROI postcontrast administration, the mean postcontrast ROI was 334 HU (standard deviation [SD] = 67 HU; p  < 10 ⁻¹⁰ ; Fig. 1B, C ).

Intraoperative Image Guidance

Intraoperative CBCT allowed preresection images and immediate postresection images to be available to the skull base surgeon ( Fig. 3 ). Surgical navigation was performed in 20 of the 21 endoscopic transnasal procedures, over which the mean (standard deviation) fiducial registration error was 1.8 mm (SD = 0.45 mm). An example of the image-guidance system registered to a postresection image is shown in Fig. 4 .

Fig. 3.

Intraoperative CBCT images acquired before (top row) and after [bottom row] tumor resection: ( A , B ) endoscopic partial anterior clival resection of inflammatory skull base mass; ( C , D ) endoscopic sphenoidotomy; ( E , F ) endoscopic resection of sinonasal schwannoma. CBCT, cone-beam computed tomography.

Fig. 4.

Surgical navigation within intraoperative 3D imaging. ( A ) Endoscopic image ( B ) axial and ( C ) coronal slices from intraoperative CBCT image acquired after partial resection of clivus. The navigated pointer (cyan) is shown to lie at the front face of the drilled clivus during an expanded endonasal approach to a skull base mass. 3D, three-dimensional; CBCT, cone-beam computed tomography.

Discussion

The adoption of intraoperative CBCT is becoming more widespread within Otolaryngology—Head and Neck Surgery. This study highlights the image quality of intraoperative CBCT scanning specific to advanced skull base surgery, demonstrating that intraoperative CBCT provides good anatomical detail allowing immediate assessment of complex resections.

Image Quality

Intraoperative imaging has been shown to increase the accuracy of surgery at the skull base, and also improve the completeness of surgical resection. ^{1 10} Two key factors for intraoperative imaging are the quality of images obtained adequate and the images can be successfully used intraoperatively to update the surgical navigation system. This study highlights that bony detail on intraoperative CBCT is excellent, with 20 key bony anatomical landmarks being scored very highly on the CBCT images. Previously the use of CBCT has been limited due to the low-contrast resolution in soft tissue imaging; limited reports in the literature demonstrate the use of IV contrast during CBCT acquisition to enhance intra-abdominal tumors and vascular lesions. ^{16 17} Despite the limited soft tissue contrast reported, the spatial resolution of CBCT is superior to multislice CT, suggesting the potential to achieve improved visualization of soft tissue and vascular structures with the use of contrast medium. ^{18 19 20} In this study, we did not see improvement of soft tissue structures (V2, V3, and optic chiasm) with the addition of IV contrast. This may be related to the IV administration technique we used, a bolus method at the beginning of the scan. When assessing the use of IV contrast using CBCT in a rabbit model, Kim et al demonstrated the best contrast enhancement when they combined a bolus injection just prior to scanning along with a continuous injection during the CBCT scanning. ²¹ However, we did see improvement of vascular structures, with key vascular landmarks (carotid artery, jugular bulb, and internal maxillary artery), being enhanced with the addition of IV contrast. Furthermore, the mean region of interest in the carotid artery postcontrast was 334 HU, consistent with Groell et al, who considered attenuation values in cervical vessels of 150 to 200 HU as sufficient for vessel visualization in contrast-enhanced spiral CT of the head and neck. ²² Due to the wide range of tumor sites and pathologies in this cohort, we did not evaluate tumor visualization with IV contrast, but this is an area for further investigation. Future research will also examine how well preoperative MRI data can be registered to an intraoperative CBCT to improve soft-tissue information if feasible.

Clinical Value

The advent of the Hopkins Rod by Harold Hopkins in the 1960's greatly advanced the possibilities of endoscopic surgery. Technology has advanced significantly since then, with surgical navigation for advanced endoscopic skull base cases now being routine. The American Academy of Otolaryngology—Head and Neck Surgery has endorsed the use of intraoperative imaging in select complex skull base surgery to assist the surgeon in clarifying complex anatomy. ² Endoscopic surgery is routinely used to treat skull base pathology; the surgeon is required to have a detailed understanding of the anatomical structures in the region for surgery to be safe and effective. Image guidance for complex sinonasal and skull base surgery has been shown to improve outcomes and reduce complication rates. ²³ Currently, image guidance systems do not represent real-time changes in surgical anatomy that would highlight changes in anatomical landmarks. The operating surgeon is therefore required to update changes in the patient's anatomy mentally as the resection progresses. Having intraoperative updated imaging allows the surgeon to assess the extent of resection, which may be particularly useful in cases where the extent of bony resection is critical to the success of the operation. Updated intraoperative images are only useful if the quality is satisfactory and the navigation system can be successfully updated with minimal registration error. The mean registration error for this updated imaging was 1.8 mm (0.45 mm), marginally smaller than the maximum registration error of 2 mm, as commonly reported in the literature. ²⁴ The excellent bony detail seen on the CBCT, along with the small registration error reported, suggests that updated imaging after resection is potentially advantageous to the surgeon, allowing assessment of resection extent. The surgeon can then go on to perform further resection if required. Cases with challenging limitations of endoscopic visualization due to overlying structures or limited surgical access are another area of potential utility for this technology.

Radiation exposure to the staff and patient must be considered when performing in-room scans. In our protocol, all clinical staffs left the room during imaging, except for the imaging technologist and the anesthetist who made use of lead shielding. In this configuration, with in-room staff far from the imaging source, occupational exposure is well below regulatory limits and typical values from interventional radiology. ²⁵ For the patient, imaging studies have demonstrated that radiation doses from CBCT are low in comparison to diagnostic CT, ^{11 25} which allows for repeat (approximately 2–5) intraoperative CBCT acquisitions with less total radiation dose than a nominal head CT (approximately 2 mSv). ²⁶

Limitations of the Study

In this study, we did not compare directly the quality of CBCT images compared with the preoperative CT scans, although it is well reported that conventional CT scanning is superior to CBCT in terms of image quality. Furthermore, we did not assess directly the additional time that each scan takes, but approximately 5 to 10 minutes for each intraoperative update is a reasonable estimate. The added time for a scan, must be balanced against the value that the scan gives to the operating surgeon. Only if the image quality is sufficient, and the scan provides new, useful information, is there value in performing intraoperative CBCT scanning.

Conclusion

Advances in intraoperative CBCT imaging have the potential to facilitate precision surgery at the skull base, allowing accurate resection of pathology while preserving the critical structures that make skull base surgery so complex.

Funding Statement

Funding This work was supported by research funding from Siemens Healthcare, the AHSC AFP Innovation Fund (Ontario Ministry of Health and Long Term Care), and the Princess Margaret Cancer Foundation.