Eur Arch Otorhinolaryngol (1997) 254:343-346 W. F r e y s i n g e r 9A. R. G u n k e l 9 Springer-Verlag 1997 9W. F. T h u m f a r t image-guided endoscopic ENT surgery Received: 5 August 1996 / Accepted: 19 December 1996 A b s t r a c t Intraoperative three-dimensional (3D) "navigation" has siginificantly improved patient safety during operative procedures on the paranasal sinuses and the frontal skull base. The ISG Viewing Wand (ISG Technologies, Mississauga, Ontario, Canada) is now used routinely for such procedures in our hospital at the present time. Current use with our radiological and intraoperative protocols has demonstrated a clinical accuracy o f 1-2 mm. Initial experience with the A R T M A Virtual Patient ( A R T M A Biomedical, Vienna, Austria) has allowed endoscopic 3D navigation by using augmented reality techniques and has been found to be very promising. We present our experience with these systems and discuss the impact of these unique techniques on computer-assisted surgery (CAS) in otorhinolaryngology and head and neck surgery. fit from such systems by extending the limits of a "safe" area to allow for more complete and radical operative therapy. Trainees and y o u n g e r surgeons m a y benefit by providing ongoing orientations to anatomic landmarks and by being warned about delicate structures that should not be touched. Recent advances in computer and software technology have allowed the development and widespread use o f intraoperative navigation systems. In contrast to neurosurgery, this field is still very new for E N T surgeons. This report will present our experience with two systems in performing surgery o f the paranasal sinuses and anterior skull base. Materials and methods Computer-assisted surgery 9ISG Viewing Wand 9A u g m e n t e d reality Intraoperative navigation 9 Image-guided surgery Key words Introduction Currently available radiological imaging provides the necessary resolution and quality for intraoperative navigation. The generation of so-called three-dimensional "3D" data, i.e., cubic voxels with contiguous slices, is necessary. Helical axial scans of the head from the upper incisors to the frontal sinus are performed to yield 1-mm slices and are used for actual navigation [4]. In this technique, 3D magnetic resonance (MR) sequences [4] are transferred entirely via the hospital network. Post-processing on a graphic workstation (Allegro, ISG Technologies, Mississauga, Ontario, Canada) generates a 3D reconstruction of the patient, which is then copied to the Viewing Wand in the operating room. A final check of the navigation unit will complete preparations for surgery within 24 h. When we started "navigation" in our hospital, we exclusively used radiopaque markers ("X-spots" Beekly Corp, Conn., USA) which were placed on the patient's head for CT scanning. We now use almost exclusively our registration device [3], which is essentially based on an upper jaw imprint, and is securely held in place by vacuum. Patient immobilization during scanning, if necessary, is achieved with a dedicated head-fixation device [9]; patients are instructed to keep their eyes shut. Similar to neurosurgical interventions, surgical procedures in otorhinolaryngology are ideally suited to the use of intraoperative computer-assisted stereotactic navigation systems. The surgical field is very precisely contained in b o n y structures, so that preoperative imaging will match exactly any perioperative findings. In both disciplines, a variety of vital structures is located in the surgical field. Especially in the presence o f changed anatomical relationships due to pathological processes or bleeding, intraoperative orientation and guidance represent an important, if not the most important, supplementary aid for a surgeon. Even very experienced surgeons can bene- The Viewing Wand System and its intraoperative use W. Freysinger (N~). A. R. Gunkel . W. F. Thumfart ENT Department, University of [nnsbruck, Anichstrasse 35, A-6020 Innsbruck, Austria The patient's 3D coordinates are determined with a position-sensitive, articulated mechanical arm, that has 6 degrees of freedom. A variety of probes (long, short, and bent probes, and straight suction tubes) is available for navigation. The intraoperative position of a probe is displayed in axial, coronal, sagittal, and "ultrasound" views of the 3D data set, and in the patient's 3D reconstruction. 344 More technical details about the navigation system can be found elsewhere [11]. After introduction of general anesthesia, the patient's head is placed in a Vogele-Bale-Hohner (VBH) head holder; underpressure repositioning of the mouthpiece is started and counterfixation is mounted. Hydraulic arms hold the components and facilitate efficient patient immobilization. Registration is performed by touching markers, the registration device, or anatomical landmarks with the calibrated probe attached to the mechanical arm and subsequent correlation to the appropriate locations in the 3D data set (Fig. 1). System preparation is completed by verification of registration [3, 4] and surgery is conducted as usual. A The Artma Virtual Patient Fig. 1 Schematic setup of the ISG navigation system for a video endoscopic procedure. Video equipment (A), computer (B), position sensitive arm (arrow). Fixation fo the patients's head to the table is shown symbolically (hatched structure). The video-endoscopy unit is not linked to the navigation system A B Fig.2 Schematic setup of the A R T M A System. The video endoscopic equipment (A) is connected to the navigation system (B), the 3D-digitizer consists of the digitizer itself (C), Hall sensors and the generator of the magnetic field (arrows) In addition to 3D CT/MR images, this system allows 3D navigation in 2D medical images, such as X-rays, ultrasound and video. This is realized by linking a 3D reconstruction of the imaged scene [5] to CT/MR, or a camera simulation [8] to overlay stereotactic targets, structures and access pathways onto available images and live video available in the computer [1]. Most importantly, this system does not need mechanical fixation of the patient on the operating table. Once the patient is anesthetized, Hall sensors are attached to the patient's head, an endoscope and the instrument. For navigation, the endoscope's optical system has to be calibrated. [6] At least six anatomic landmarks, the registration device or fiducials are touched with a stylus and correlated to a specific location in each of the navigable images [10]. CT-MR registration is performed as usual [11]. Visual inspection of the concordance of calculated and measured points on the live video terminates the system's setup. Surgery is carried out as usual. Prior to the actual intervention the CAS operator and the surgeon jointly define the structures to be visualized in the live video, such as the optimum access trajectory to a target or a vital anatomical structure to avoid (Fig. 2) Results T h u s far, w e h a v e u s e d t h e I S G V i e w i n g W a n d a n d t h e ARTMA Virtual Patient in a series of 79 interventions ( T a b l e 1). B o t h C T a n d M R d a t a h a v e b e e n u s e d s u c c e s s - T a b l e 1 Summary of indications for 3D navigational surgery Indication for surgery Operations System (Pan-)sinusitis 20 ISG ARTMA 15 5 3 3 FESS 25 ISG ARTMA 20 5 3 3 5 ISG ARTMA 5 1 < 1.5 n/a Samples from the maxillary sinus, the nasal septum and the sinus Morgagni from the "Innsbruck Iceman" [6] Orbital decompression 10 ISG ARTMA 8 2 < 1.5 3 For intraoperative accuracies approaching the monitor resolution < 1.5 mm, arbitrarily chosen, resembles the accuracy of a surgeon to quantify the position of an instrument video-endoscopically Carcinoma (primary and recurrent adenocarcinoma, adenoid cystic carcinoma) 15 ISG ARTMA 13 2 2 3 4 ISG ARTMA 1 3 2 3 Biopsies (posterior sphenoid wall [2], roof of sphenoid, pterygopalatine fossa, posterior maxillary sinus) Foreign bodies [aspergillus sinusitis (2 cases) dental filling in maxillary sinus, ingested needle in parapharyngeal space] Mean accu- Remarks racy (ram) Tape and VBH head fixation [4] estimated accuracy Excessive intraoperative tissue shifting of the neck did not allow removal 345 Fig.3 Intraoperative screenshot of a biopsy facilitated by the Viewing Wand in the sphenoid sinus. It shows axial, coronal and sagittal, and the ultrasound reformatted CT views (counter-clockwise). The tip of the Viewing Wand's probe is located at the center of the crosshairs, which is positioned on the bottom of the sphenoid. The contrast-enhanced CT shows the relation of the surgical field to the internal carotid arteries. The tumor is seen eroding the apex of the left orbit Fig. 4 [ntraoperative video screenshot during the removal of a foreign body with the ARTMA Virtual Patient system: navigation employs X-ray, CT, and video images. The planned path into the maxillary sinus (rectangular structures), the target (point) and local coordinate systems in the static view are shown fully for navigation. The clinical data for arm-based intraoperative localization have been more detailed to date. W h e n using the V B H headholder and the appropriate registration techniques we have typically reached acccuracies in the range o f the monitor's resolution. The following cases exemplify use of both 3D systems. One patient, 29 years/male, had extensive neurological deficits due to tumor and required a transnasal, transsphenoidal biopsy of the posterior wall of the sphenoid sinus to histologically classify the tumor lesion present. A n intraoperative registration accuracy of < 2 m m in all directions allowed us to approach the tumor close to the inter- Fig. 5 Intraoperative screenshot of the surgery in Fig. 4. A photograph of the live endoscopic endonasal video. The endoscope is just being inserted through the nostril. The right concha (left structure), a nasal spur (lower right structure), some polypoid structures (upper center) can be differentiated. The endoscope is guided through the rectangles, symbolizing a bent 3D path, which curves at the uncinate process (point) to the predefined target hal carotid arteries (Fig. 3). Navigating with the I S G system's probe confirmed the exact anatomic location of the arteries so that the procedure was completed without complications. Surgery was completed without complications. The biopsy revealed an adenoid cystic carcinoma, which was subsequently resected by neurosurgical staff. The second patient, 32 years/female, reported headache with obstructed right nasal breathing and discharge. Imaging revealed a radiodense structure in the right maxillary sinus and thickened endonasal m u c o s a (Fig. 4). A foreign b o d y was located in the right maxillary sinus by the A R T M A system. Planning was executed on CT scans and a lateral X-ray of the patient. Intraoperatively, a nasal antral fenestra was created and a = 2 x 3 x 3 m m amalgam dental filling was successfully removed. This was extensively covered with aspergillus. Surgery was completed without complications. In this case, the introperative accuracy was 3 mm. Discussion In our routine work with the Viewing Wand, we have achieved mechanical and registration accuracies of 0.2 and < 2 mm, respectively. The registration device [9] allows minimizing intraoperative errors to the resolution of the Viewing Wand's monitor. For the A R T M A system we have found accuracies better than 3 mm. This number has to be estimated, since the intraoperative videoendoscopic images superimposed with stereotactic information have to be evaluated. The Virtual Patient technology was found to be very promising. The interferences of magnetizable objects with the magnetic digitizing technology, however, can be han- 346 died satisfactorily [7]. The system has advanced real-time on-line telesurgery capabilities based on telephone, ISDN, Ethemet, and Internet [8], as demonstrated by sharing a CT data set, registration parameters, and the surgical planning of revision sinus surgery between Innsbruck, Austria and San Diego, California, USA, via ISDN [7]. The two systems studied represent two different stages of the technical evolution in CAS technology. The ISG System is widely used in neurosurgery and has already gained significant attention in ENT surgery [2, 10], whereas our department has been the first clinical otorhinolaryngologic test site for the ARTMA system. Both systems provide comparable intraoperative accuracy, although the ISG system to date has performed slightly better. However, this is more than compensated by the limitations imposed by the technologic approach chosen for navigation. The measurement of position needs a dedicated instrument, and the handling of a surgical instrument, if mounted to the articulated arm, is impeded by the mechanical limitations and the weight imposed by the arm. We believe the development of detailed application protocols will allow both 3D navigation systems to achieve similar accuracies. A comparison of two such different techniques still needs to be done with caution. The Virtual Patient system is designed to integrate naturally into a video endoscopic procedure without "disturbing" an intraoperative setup by mechanical fixation of a patient or mechanical arms introduced in the sterile zone. This represents a state-of-the-art solution for providing real-time positional information to the surgeon in a realtime endoscopic video. It can be anticipated that the clinical performance of augmented reality in the operating room will have a substantial impact on the development of future intraoperative navigation systems. The use of high-tech navigation in our hands has not resulted in intraoperative complications induced by system usage. However, the procedures safely performed so far in highly critical areas (such as in the first patient described) or in facilitating tumor resections suggest the real value of this technique. Patient safety can be and was maximized by use of these two navigation systems. References 1. Abdel-Aziz YI, Karara HM (1971) Direct linear transformation from comparator coordinates into object coordinates. In: Proceedings of the Symposium on Close-Range Photogrammetry Urbana, Ill, USA, January 1971 2. Anon J, Rontal M, Zinreich SJ (1995) Computer-assisted endoscopic sinus surgery - current experience and future developments. Operative techniques. Otolaryngol Head Neck Surg 6 : 163-170 3.Bale RJ, Vogele M, Freysinger W, Gunkel AR, Martin A, Bumm K, Thumfart WF (1997) A minimally invasive head holder to improve the performance of frameless stereotactic surgery. Laryngoscope 107:373-377 4. Freysinger W, Gunkel AR, Vogele M, Bale RJ, Thumfart WF (1996) Einsatz der ISG Viewing Wand in der HNO-Chirurgie. ORL Nova 6 : 223-228 5. Longuet-Higgins HC (t 981) A computer algorithm for reconstruction a scene from two projections. Nature 293 : 133 6. Thumfart WF, Freysinger W, Gunkel AR (1996) 3D imageguided surgery on the example of the 5300 year old Innsbruck Iceman. Acta Otolaryngol 117:131-134 7. Truppe MJ, Pongracz F, Mayrbaeurl J, Freysinger W, Gunkel AR, Thumfart WF (1995) Augmented reality remote-guided navigation in ENT surgery. Medicine Meets Virtual Reality 4, 19-22 January 1995, San Diego, Calif, USA 8. Truppe M, Pongracz F, Freysinger W, Gunkel AR, Thumfart WF (1995) Interventional video tomography. In: Lemke HU, Inamura K, Jaffe CC, Vannier MW (eds) CAR '95. Proceedings of the International Symposium on Computer and Communication Systems for Image Guided Diagnostics and Therapy, Springer, Berlin, p 1303 9. Vogele M, Bale RJ, Martin A, Freysinger W, Gnnkel AR, Thumfart WF (1994) A new registration device for stereotactic ENT surgery. Annual Meeting of the West German ENT Society, 19-20 April 1994, Aachen, Germany 10.Vogele M, Bale R, Freysinger W, Thumfart WF (1996) Kann die ISG Bildprtifwand ftir Operationen am Felsenbein verwendet werden? ORL Nova 6 : 38 11. Zinreich SJ, Tebos SA, Long DM, Brem H, Mattox DE, Loury ME, Vander Kolk CE, Koch WM, Kennedy DW, Bryan RN (1993) Frameless stereotactic integration of CT imaging data: accuracy and initial applications. Radiology 188:735-742