Strahlentherapie und Onkologie
Original Article
Multimodality Cranial Image Fusion Using External Markers Applied via a Vacuum Mouthpiece and a Case Report* Reinhart A. Sweeney1, Reto J. Bale2, Roy Moncayo3, Karl Seydl1, Thomas Trieb2, Wilhelm Eisner4, Johannes Burtscher4, Eveline Donnemiller3, Günther Stockhammer5, Peter Lukas1
Purpose: To present a simple and precise method of combining functional information of cranial SPECT and PET images with CT and MRI, in any combination. Material and Methods: Imaging is performed with a hockey mask-like reference frame with image modality-specific markers in precisely defined positions. This frame is reproducibly connected to the VBH vacuum mouthpiece, granting objectively identical repositioning of the frame with respect to the cranium. Using these markers, the desired 3-D imaging modalities can then be manually or automatically registered. This information can be used for diagnosis, treatment planning, and evaluation of follow-up, while the same vacuum mouthpiece allows precisely reproducible stereotactic head fixation during radiotherapy. Results: 244 CT and MR data sets of 49 patients were registered to a root square mean error (RSME) of 0.9 mm (mean). 64 SPECTCT fusions on 18 of these patients gave an RMSE of 1.4 mm, and 40 PET-CT data sets of eight patients were registered to 1.3 mm. An example of the method is given by means of a case report of a 52-year-old patient with bilateral optic nerve meningioma. Conclusion: This technique is a simple, objective and accurate registration tool to combine diagnosis, treatment planning, treatment, and follow-up, all via an individualized vacuum mouthpiece. Especially for low-resolution PET and even more so for some very diffuse SPECT data sets, activity can now be accurately correlated to anatomic structures. Key Words: Image registration · Image fusion · Radiotherapy · Multimodality · Head fixation · PET · SPECT Strahlenther Onkol 2003;179:254–60 DOI 10.1007/s00066-003-1031-2 Multimodale Bildfusion anhand externer Referenzmarker über ein Vakuummundstück in exakter Relation zum Schädel, demonstriert an einem Fallbeispiel Ziel: Vorstellung einer einfachen und präzisen Methode zur Korrelation nuklearmedizinischer Bildgebung (SPECT und PET) des Schädelbereichs mit CT und/oder MRT. Material und Methodik: Die Bildgebung erfolgt mit einem externen helmartigen Referenzrahmen mit integrierten modalitätsspezifischen Markern an definierten Positionen. Die genaue Relation des Rahmens zum Schädel ist durch reproduzierbares Aufstecken auf das VBH-Vakuummundstück gewährleistet. Mittels dieser Marker können die jeweiligen Datensätze manuell oder (halb)automatisch registriert werden. Die so fusionierten Bilder können in der Diagnostik, Bestrahlungsplanung und Nachsorge eingesetzt werden; das Vakuummundstück dient gleichzeitig der präzisen Lagerung bei fraktionierter stereotaktischer Bestrahlung. Ergebnisse: 244 CT- und MRT-Datensätze von 49 Patienten wurden mit einem durchschnittlichen „root square mean error“ (RSME) von 0,9 mm überlagert, während an 18 Patienten 64 CT-SPECT-Fusionen zu einem RSME von 1,4 und 40 CT-PET-Datensätze von acht Patienten zu 1,3 mm fusioniert wurden. An einer 52-jährigen Patientin mit bilateralem Optikusscheidenmeningiom wird die Methode beispielhaft demonstriert. Schlussfolgerung: Diese Methode ist ein einfaches, objektives und genaues Verfahren zur Registrierung multimodaler Datensätze. Insbesondere die Integration von Datensätzen mit niedriger Auflösung (PET und SPECT) erlaubt bedeutende Einblicke in klinische Vorgänge, die teilweise aus CT bzw. MRT nicht ersichtlich sind. Schlüsselwörter: Radiotherapie · Bildfusion · SPECT · PET · Kopffixation · Strahlentherapie · Multimodal 1
Department of Radiotherapy-Radiooncology, Department of Radiology 1, Interdisciplinary Stereotactic Intervention- and Planning Laboratory (SIP-LAB), 3 Department of Nuclear Medicine, 4 Department of Neurosurgery, and 5 Department of Neurology, University Hospital Innsbruck, Austria. 2
*Presented at: DEGRO/ÖGRO/DGMP, Munich, Germany, October 2000 (Elekta Poster Award for functional imaging and treatment planning), ASTRO 2001, San Francisco, CA, USA, and RSNA 2001, Chicago, IL, USA. Received: April 8, 2002; accepted: October 21, 2002
254
Strahlenther Onkol 2003 · No. 4 © Urban & Vogel
Sweeney RA, et al. Multimodality Cranial Image Fusion
Introduction It is now general practice to incorporate both CT and MRI images into the treatment-planning process, since CT images, although to date largely inexpendable due to their role in dose planning and excellent spatial resolution, will often not provide adequate imaging of critical soft tissue structures such as the optic chiasm or tumors themselves. MR images on the other hand, although subject to distortion artifacts [8], often provide the required resolution leading to improvements in target volume definition [23]. The functional information offered by SPECT/PET images can also improve understanding of tumor size, location and activity, possibly also leading to changes of actual treatment volumes [9–11]. A general problem, however, when comparing any two sets of one patient’s images without prior fusion, is that these studies are generally not performed with rigorous attention paid to identical patient positioning or scanner protocols. Thus, the slices are rarely in the same anatomic plane. In addition, differences in scanner characteristics and protocols (slice thickness, spacing, pixel size, matrix size, etc.) hardly allow correlation of features in separate scans with 3-D accuracy < 5 mm [21]. We present the application of a noninvasive, simple and fast method of image registration allowing precise and objective assignment of functional information from SPECT and/or PET to CT/MRI, based on external landmarks as reference points. Their connection to the rigid upper palate via the same vacuum mouthpiece, used for noninvasive head fixation in (stereotactic) radiotherapy [25], guarantees their identical relationship to the cranial anatomy.
In short, spherical external markers can be placed in precise relation to cranial anatomy and made visible in all imaging modalities via a hockey mask-like frame, in contact with the cranium only via a vacuum mouthpiece (Figure 1). Geometrically identical frames, with modality-specific markers (Table 1) screwed into precisely defined positions on the frames, are lo-
Material and Methods The system, consisting of the SIP-Lab Innsbruck frame, reproducibly connected to the Vogele-Bale-Hohner (VBH) vacuum mouthpiece (both Medical Intelligence GmbH, Schwabmünchen, Germany), and the methodology of image registration have been previously described [3].
Abbildung 1. Patient vor Beginn einer SPECT-Aufnahme mit dem „SIPLab Innsbruck frame“. Dieser befindet sich dank eines Vakuumzahnabdrucks (1), der über einen Schlauch (2) an eine Vakuumpumpe angeschlossen ist (nicht sichtbar), in exakter Relation zum Schädel. Das Vakuum baut sich nur bei absolut korrektem Sitz des Zahnabdrucks auf. Zur Bildregistrierung werden mindestens vier (bis zu zwölf) Referenzmarker (3) jedes Datensatzes miteinander korreliert.
Figure 1. Patient in scanner with SIP-Lab Innsbruck frame reproducibly attached to a vacuum mouthpiece (1). A vacuum pump (not visible), connected via a vacuum hose (2) to this mouthpiece creates the required negative pressure which only builds if the mouthpiece is seated against the upper palate perfectly. Thus, at least four (up to twelve) reference points (3) are in precise relation to cranial anatomy and, since they are visible in all modalities, can then be correlated to each other in the registration process.
Table 1. Acquisition protocols and scanner specifics. Tabelle 1. Scannerdaten und relevante Akquisitionsprotokolle. CT SIP frame marker Matrix FOV (mm) Slice thickness (mm) Voxel size (mm) Scanner Other
Spherical glass beads 512 512 250 2.5 0.5 0.5 2.5 GE Lightspeed
MRI
SPECT
PET
Nitrolingual capsules 512 512 260 2.5 0.5 0.5 2.5 Siemens Magnetom Vision 1.5 T, head coil
241
Spherical glass beads
Strahlenther Onkol 2003 · No. 4 © Urban & Vogel
Am
64 64 326 5.1 5.1 5.1 5.1 ADAC Vertex Two separate energy windows (patient- and frame-tracer) Reconstruction and manipulation with ADAC PEGASYS
128 128 550 4.25 4.3 4.3 4.25 GE-Advance Frame markers from transmission scan
255
Sweeney RA, et al. Multimodality Cranial Image Fusion
cated at each respective scanner. Specifics on data acquisition protocols are also listed in Table 1. With the help of commercially available software (“Pairpoint matching” Framelink/Cranial; Stealth navigation sysTable 2. Number of registrations between the different modalities, their mean RSME, and the range of registration errors. Tabelle 2. Anzahl der Registrierungen der unterschiedlichen Bildmodalitäten, ihr mittlerer RSME und Spannbreite der Ergebnisse.
n Mean RSME (mm) Range (mm)
CT-MR
CT-SPECT
CT-PET
244 0.9 0.2–1.9
53 1.4 0.4–1.9
16 1.6 0.4–1.9
tem, Medtronic, Boulder, CO, USA), these images were manually registered according to these markers. When manually registering two data sets, registration accuracy is calculated by the software as the root square mean error (RSME) which is the mean distance of the respective frame reference points in the two data sets. The upper tolerance level of RSME was preset to 2 mm. Obviously, this RSME itself says nothing on the actual anatomic registration accuracy. This assumption is, however, granted by the fact that the frame is reproducibly connected to the vacuum mouthpiece, which itself is in reliable relation to the cranium thanks to the negative pressure acting on the hard palate [18]. Up to ten further previously registered studies, be it CT, MRI or functional imaging studies (SPECT/PET), can then be uploaded and compared to each other individually. The threshold levels of all image sets as well as the weighting of one modality over the other can be adjusted via mouse-controlled sliders, allowing quick visualization of any region in any magnification in three planes (axial, coronal, sagittal).
Abbildung 2. Bildschirmansicht des CT vor der Bestrahlung (2,5 mm Schichtdicke). Am rechten Sehnerv (seit Jahren erblindet) zeigt sich eine auffällige Verdickung, der linke Sehnerv, obwohl histologisch befallen und klinisch fast erblindet, ist unauffällig (Pfeil in der axialen Ansicht). Im unteren rechten Quadranten ist eine 3-D-Rekonstruktion der Patientin mit links frontaler Kraniotomie und Referenzmarkern des Rahmens sichtbar.
Results Vacuum mouthpiece production, as a simple, one-time procedure, takes 10–20 min. Inside the scanner room, hose connection to the (portable) vacuum pump, vacuum mouthpiece insertion, and mounting of the SIP frame adds an additional 1–2 min to image acquisition time. It must be emphasized that there is no need for either reproducible positioning of the head or head fixation in any scanner. However, for uncooperative patients and during longer acquisition periods (especially SPECT), fixation is possible via mechanical arms to reduce motion artifacts. Since these arms are not MRcompatible, we routinely tape the protruding anterior extensions of the mouthpiece to the head coil to reduce motion artifacts, with satisfactory results. Registration, in this study still performed manually, requires between 4 and 10 min per data pair. Using this method, correlation of functional with anatomic data is routinely performed at our hospital where to date, well over 100 patients have received radiation treatment of cranial tumors fixated via the vacuum mouthpiece. CT/MRI fusion was performed in 49 of these, SPECT or PET information was incorporated in 26 patients. This information was used not
256
Strahlenther Onkol 2003 · No. 4 © Urban & Vogel
Figure 2. Screencapture of CT before radiotherapy (2.5 mm slice thickness). Note thickening of right optic nerve (irrreversibly blind) but inconspicuous left optic nerve (arrow in the axial view) at this point < 10% vision. Note also the left frontal craniotomy and fiducial markers of the frame on the 3-D reconstructed view in the lower right quadrant.
Sweeney RA, et al. Multimodality Cranial Image Fusion
Figure 3a. Screencapture of pre-irradiation CT-99mTc-octreotide SPECT fusion; note intense activity over right and left optic nerve. Also visible are two of the (CT and SPECT) reference markers on the frame. Abbildung 3a. Bildschirmansicht der prätherapeutischen CT-99mTc-Octreotid-SPECT-Fusion; intensive Aktivität nicht nur über dem rechten, sondern auch über dem linken Auge. Ebenfalls sichtbar sind zwei CT- und SPECT-Referenzmarker am Rahmen.
only in radiotherapy treatment planning and followup (as required, i.e., when unclear whether tumor growth or different slicing/active tumor vs necrosis) but also, if available, in the preceding diagnostic procedures and neurosurgical interventions [1]. If all follow-up studies are included, a total of 311 registrations were performed at our department with this method to date (mean of 6.7 registrations per patient). The mean RSMEs of these registrations are shown in Table 2.
Figure 3a
Case Report One of the many interesting cases where the coregistration gave insight into events which would have otherwise gone unnoticed is recounted below. A 52-year-old woman presented with a 2-year history of complete blindness on the right and rapidly progressing loss of vision of the left eye. Initial imaging showed only thickening of the right optic nerve. A right temporal surgical approach exposing both optic nerves to the chiasm showed an atrophic right optic nerve and an uninvolved chiasm. Biopsy of the visible tumor on the right and the arachnoid sheath of the left optic nerve revealed a bilateral psammomatous meningioma. At initial presentation to our department, only light/dark could be distinguished with the left eye. A vacuum mouthpiece was made at the initial consult. The SIP frame with the respective modality-specific markers was then reproducibly attached to this vacuum mouthpiece for (planning) CT, MR and SPECT imaging. The RSME of the CT/SPECT images was 1.25 mm. CT and MRI both showed enlargement of
Figure 3b. The identical slice 4 weeks after radiotherapy; no more abnormal activity over left optic nerve (correlating with clinical response of regained [80%] vision), while that over the right optic nerve (outside the 50% isodose line) remained unchanged.
Figure 3b
Strahlenther Onkol 2003 · No. 4 © Urban & Vogel
Abbildung 3b. Die gleiche Schicht, 4 Wochen nach Bestrahlung; links retrobulbär nun keine abnorme Aktivität mehr (Visus ca. 80%), während die retrobulbäre Aktivität rechts unverändert geblieben ist (außerhalb der 50%-Isodose).
257
Sweeney RA, et al. Multimodality Cranial Image Fusion
the right optic nerve, but the left optic nerve, in contrast to the clinical situation, was inconspicuous (Figure 2). 99mTc-tricineHYNIC-D-Phe1-Tyr3-octreotide SPECT (4 h post-injection) [6] superposition then confirmed the diagnosis of bilateral meningioma, showing intense somatostatin receptor activity over the right and, to a slightly lesser degree, the left optic nerve (Figure 3a). The intraorbital aspect of the left optic nerve up to the optic chiasm was delineated as gross tumor volume (GTV), with 5 mm added as planning treatment volume (PTV) as defined by the International Commission on Radiation Units and Measurements [15]. 50 Gy to the 95% prescription isodose was applied at 1.8 Gy/d via multiple noncoplanar fields. The same individualized mouthpiece served as point of reproducible fixation for the actual treatment [25]. After week 2 of treatment, vision of the left eye had markedly improved so that she could again read the newspaper. Much to her delight, vision continued to improve to 80% 3 weeks after treatment, still stable 3 years later. While followup CT and MRI (again with vacuum mouthpiece and SIP frame) showed no discernible difference to the pretreatment status, 99mTc-octreotide SPECT showed an impressive response to treatment with no more abnormal activity in the left retrobulbar region (GTV) corresponding to the clinical improvement (Figure 3b). The right retrobulbar activity (outside the PTV), on the other hand, remained virtually unchanged (the activity seen in the pituitary and nasal cavity regions is physiological [20]). Further follow-up studies will continue to be performed with the same vacuum mouthpiece and frame since this adds only minimal additional time to the imaging procedure.
take can be a guessing game. A reliable method of registration with CT or MRI to accurately define localization would be of high diagnostic utility. While a detailed discussion of all fusion methods far exceeds the scope of this article, excellent review articles and books on medical image registration are available [12, 17]. In short, all modalities must be brought into spatial alignment (registration). Then the fusion, that is the integrated display of involved data, can be performed. Table 3 gives a general overview of the different methods, usually based on either intrinsic (anatomic) features, extrinsic features or non-imagebased (calibrated coordinate) systems. Accuracy using contours for registration (such as the “head and hat” [21] or the “iterative closest point” algorithm) may yield gross misregistrations, although the contours align perfectly due to identical axes of symmetry. Different imaging modalities can also provide substantially different image contrast between corresponding surfaces [13]. Methods based on intrinsic (anatomic) landmarks pose the challenge of finding three or more appropriate, precisely definable landmarks, requiring skill and practice of the user. Our own experience is in accordance with that of other groups [7] in that such anatomic landmarks can be coregistered to about 2–5 mm for CT and MRI. Such landmarks are, however, hard to define in SPECT/ PET images. External reference points or stereotactic frames, however, if reliably in identical relationship to cranial anatomy, offer this possibility. This identical relationship can be granted by invasive frames, implanted bone markers, thermoplastic masks [14], skin markers or, as in ours, via a noninvasive frame reproducibly attached to a vacuum mouthpiece. Discussion Invasive frames, of course, are limited by their short-term In many clinical situations, the additional information and aconly application. Thus, fusion of follow-up images and, should curacy gained by image registration and fusion can be of great fixation for radiotherapy be required, the option of fractionadvantage in the treatment-planning process. Although excelated treatment are hardly possible. An invasive procedure lent algorithms for CT and MRI fusion exist, attaining similar solely for the purpose of image fusion is, in our opinion, not acaccuracy and objectivity with the lower resolution PET and ceptable or necessary for that matter. especially SPECT image registration remains a challenge. EsWhen using skin markers, skin shift was the factor causing pecially for some types of functional tracers (i.e., 201Tl chlomost inaccuracies (errors in the direction perpendicular to the ride) where anatomic structures (such as cranial contours) are skin surface were generally much smaller than those in other hardly discernible, assigning an anatomic correlate to focal updirections) [22]. In addition, the lack of consistency over a longer period (weeks to years) makes this option less attractive. Attaching Table 3. A brief outline of commonly used methods for image registration. markers or stereotactic frames to maskTabelle 3. Übersicht der gebräuchlichsten Methoden der Bildregistrierung. based fixation systems is an interesting and viable method if one assumes high Intrinsic Landmark-based Anatomic/geometric repositioning accuracy of the underlying Segmentation-based a) Rigid models (points, curves, or surfaces) anatomy. The accuracy of all mask-based b) Deformable models (snakes/nets) systems is, however, limited by possible Voxel property-based Reduction to scalars/vectors (moments, principle axes) distortions of the underlying skin. The Extrinsic Invasive Stereotactic frame/screw markers patient would also need to be fixated for Noninvasive Thermoplastic mask/dental adapter (in combinaeach imaging procedure. tion with a stereotactic frame or directly attached The SIP-Lab Innsbruck frame with fiducials), skin fiducials its twelve markers is always and objec-
258
Strahlenther Onkol 2003 · No. 4 © Urban & Vogel
Sweeney RA, et al. Multimodality Cranial Image Fusion
tively in identical relationship to the cranium due to the vacuum of the mouthpiece, irrespective of patient compliance. Identical positioning or fixation of the head with respect to the imaging plane is not necessary, as the markers define orientation in the 3-D data set. Only if structures in the neck region are to be accurately registered between images is identical positioning and fixation of the head and neck necessary. This can be and has been performed in a variety of clinical situations at our institution with an individually formed vacuum mattress rigidly connected to the base plate of the head fixation system, in addition to the SIP frame as described. The reason the SIP frame curves around the head, thus spreading the markers as evenly as possible over a large surface with the most posterior markers located behind the ear (Figure 1), is to minimize the distance from the marker centroid to a point of interest (isocenter) in the posterior (occipital) regions of the head. This is of relevance, because the localization error increases not only as a function of marker/fiducial localization error (RSME) but also as the distance from the marker centroid to the point of interest increases [19, 26]. The frame also fits into the head coil of MR scanners, thus decreasing effects of image distortions [8]. Since the markers are larger than the dimensions of a single voxel, defining their center of mass in large magnification allows the RSME to be well under the voxel dimension of the respective imaging modality [16]. It must be reminded, however, that the RSMEs achieved in this study only represent the mean difference of the reference points (frame) to each other between the two fused data sets, not the anatomic accuracy of the registration. Our experiences on anatomic correlation between data sets, however, confirmed the results of a phantom study conducted at this institution [4], whereby fiducial (frame) and actual target registration error between CT/MR and CT/SPECT datasets correlated to < 1.5 mm. Nonetheless, each registration should be thoroughly checked in three planes by a physician before incorporating it in the treatmentplanning process. The limitation of this system is that edentulous patients are neither ideal candidates for fusion nor fixation during the radiation treatment using the vacuum mouthpiece. While the vacuum mouthpiece does hold the SIP-Lab frame’s weight on an edentulous upper palate, accuracy of registration and repositioning is somewhat less reliable due mainly to a possible rotation around a craniocaudal axis of the vacuum mouthpiece without the patient or staff noticing, i.e., without the vacuum breaking (unpublished data by one of the authors, R.A.S.). Although the fused images of the patient described as case report, as well as the others involved to date, were used and considered for CTV volume delineation, incorporation of the automatic registration software in the actual treatment-planning software is currently in progress at our institution. In addition, the integration of software for (semi)automated marker detection has already been developed [5] and used (TIANI, Medgraph, Vienna, Austria) to expedite and simplify the reg-
Strahlenther Onkol 2003 · No. 4 © Urban & Vogel
istration process. Manual “paired-point” matching is, however, available on most modern planning systems, making this method simple to implement without costly software additions. Ideally, the vacuum mouthpiece (its production takes about 10 min requiring very little training and costing about € 40.–/patient) is made at initial presentation of a patient with symptoms suspicious of a cranial tumor. The initial 3-D CT/MRI/SPECT/PET data sets can then be registered and used for frameless stereotactic neurosurgery [1] and/or radiation planning and treatment [24, 25] as well as brachytherapy applications [2], possibly reducing the need for additional scans as the patient proceeds from department to department. Conclusion When considering the multitude of fusion software already available, we see the usefulness of this method mainly for the ever increasing incorporation of low resolution images such as SPECT and PET into the ever increasing precision of radiation treatments. The negative pressure of the vacuum mouthpiece allows the most direct, noninvasive connection to the cranium, granting not only precise repositioning for fractionated treatments but now also accurate and simple integration of multimodal cranial imaging into radiotherapy treatment planning and follow-up using an external reference frame. The reference points on this frame grant completely objective registration using relatively simple and ubiquitous software, independent of a user’s capability of defining anatomic landmarks or the varying limitations of more elaborate and costly algorithms. Acknowledgment The authors would like to express their sincere gratitude to the involved radiation technicians (CT/MRI/Nuclear medicine), especially Thomas Lang and Martin Knoflach from the SIP-Lab.
References 1. Bale RJ, Burtscher J, Eisner W, et al. Computer-assisted neurosurgery by using a noninvasive vacuum-affixed dental cast that acts as a reference base: another step toward a unified approach in the treatment of brain tumors. J Neurosurg 2000;93:208–13. 2. Bale RJ, Freysinger W, Martin A, et al. First experiences with computer-assisted frameless stereotactic interstitial brachytherapy (CASIB). Strahlenther Onkol 1998;174:473–7. 3. Bale RJ, Sweeney RA, Burtscher J, et al. A new concept of a unified approach in the treatment of brain tumors. In: Lemke HU, Vannier MW, Inamura K, Farman AG, Doi K, eds. Proceedings of the 14th International Congress and Exhibition CARS 2000. Berlin: Springer, 2001:679–83. 4. Bale RJ, Sweeney RA, Moncayo R, et al. Precise and reliable CT-MR-SPECT image fusion based on a non-invasive vacuum affixed dental frame. Radiology 2001;221:Suppl:238. 5. Capek M, Wegenkittl R, Jaschke W, et al. Multimodal volume registration based on spherical markers. Conference Proceedings of the 9th International Conference in Central Europe on Computer Graphics, Visualization and Computer Vision 2001, WSCG 2001. Pilsen: University of West Bohemia, 2001:17–24. 6. Decristoforo C, Mather SJ, Cholewinski W, et al. 99mTc-EDDA/HYNIC-TOC: a new 99mTc-labelled radiopharmaceutical for imaging somatostatin receptorpositive tumours; first clinical results and intra-patient comparison with 111 In-labelled octreotide derivatives. Eur J Nucl Med 2000;27: 1318–25.
259
Sweeney RA, et al. Multimodality Cranial Image Fusion
7. Edwards PJ, Hawkes DJ, Penny GP, et al. Guiding therapeutic procedures. In: Hanjal JH, Hawkes DJ, Hill D, eds. Medical image registration. Boca Raton–London–New York–Washington DC: CRC Press, 2001:261. 8. Fransson A, Andreo P, Poetter R. Aspects of MR image distortions in radiotherapy treatment planning. Strahlenther Onkol 2001;177:59–73. 9. Grosu AL, Weber W, Feldmann HJ, et al. First experience with I-123-alphamethyl-tyrosine spect in the 3-D radiation treatment planning of brain gliomas. Int J Radiat Oncol Biol Phys 2000;47:517–26. 10. Hamilton RJ, Blend MJ, Pelizzari CA, et al. Using vascular structure for CTSPECT registration in the pelvis. J Nucl Med 1999;40:347–51. 11. Hamilton RJ, Sweeney PJ, Pelizzari CA, et al. Functional imaging in treatment planning of brain lesions. Int J Radiat Oncol Biol Phys 1997;37:181–8. 12. Hanjal JH, Hawkes DJ, Hill D, eds. Medical image registration. Biomedical Engineering Series. Boca Raton–London–New York–Washington DC: CRC Press, 2001. 13. Hawkes DJ. Registration methodology: introduction. In: Hanjal JH, Hawkes DJ, Hill D, eds. Medical image registration. Boca Raton–London– New York–Washington DC: CRC Press, 2001:25. 14. Henze M, Schuhmacher J, Hipp P, et al. PET imaging of somatostatin receptors using [68Ga] DOTA-D-Phe1-Tyr3-octreotide: first results in patients with meningiomas. J Nucl Med 2001;42:1053–6. 15. ICRU. Prescribing, recording, and reporting photon beam therapy. Report 50. Washington: ICRU, 1993. 16. Maciunas RJ, Galloway RL Jr, Latimer JW. The application accuracy of stereotactic frames. Neurosurgery 1994;35:682–94. 17. Maintz JB, Viergever MA. A survey of medical image registration. Med Image Anal 1998;2:1–36. 18. Martin A, Bale RJ, Vogele M, et al. Vogele-Bale-Hohner mouthpiece: registration device for frameless stereotactic surgery. Radiology 1998;208: 261–5. 19. Meeks SL, Bova FJ, Wagner TH, et al. Image localization for frameless stereotactic radiotherapy. Int J Radiat Oncol Biol Phys 2000;46:1291–9. 20. Moncayo R, Baldissera I, Decristoforo C, et al. Evaluation of immunological mechanisms mediating thyroid-associated ophthalmopathy by radionuclide imaging using the somatostatin analog 111In-octreotide. Thyroid 1997;7: 21–9.
260
21. Pelizzari CA, Chen GT, Spelbring DR, et al. Accurate three-dimensional registration of CT, PET, and/or MR images of the brain. J Comput Assist Tomogr 1989;13:20–6. 22. Peters AR, Muller SH, de Munck JC, et al. The accuracy of image registration for the brain and the nasopharynx using external anatomical landmarks. Phys Med Biol 2000;45:2403–16. 23. Rasch C, Keus R, Pameijer FA, et al. The potential impact of CT-MRI matching on tumor volume delineation in advanced head and neck cancer. Int J Radiat Oncol Biol Phys 1997;39:841–8. 24. Sweeney R, Bale R, Vogele M, et al. Repositioning accuracy: comparison of a noninvasive head holder with thermoplastic mask for fractionated radiotherapy and a case report. Int J Radiat Oncol Biol Phys 1998;41: 475–83. 25. Sweeney RA, Bale R, Auberger T, et al. A simple and non-invasive vacuum mouthpiece-based head fixation system for high precision radiotherapy. Strahlenther Onkol 2001;177:43–7. 26. Zinreich SJ, Tebo SA, Long DM, et al. Frameless stereotaxic integration of CT imaging data: accuracy and initial applications. Radiology 1993;188: 735–42.
Correspondence Address Dr. Reinhart A. Sweeney Department of Radiotherapy-Radiooncology University Hospital Innsbruck Anichstraße 35 6020 Innsbruck Austria Phone (+42/512) 504-2800, Fax -2812 e-mail:
[email protected]
Strahlenther Onkol 2003 · No. 4 © Urban & Vogel
