Medical Physics Letter Case report of a near medical event in stereotactic radiotherapy due to improper units of measure from a treatment planning system D. J. Gladstonea) Division of Radiation Oncology, Department of Medicine, Dartmouth-Hitchcock Medical Center, Norris Cotton Cancer Center, Lebanon, New Hampshire 03756
S. Li Division of Radiation Oncology, Department of Medicine, Dartmouth-Hitchcock Medical Center, Norris Cotton Cancer Center, Lebanon, New Hampshire 03756 and Department of Radiation Oncology, Temple University Hospital, Philadelphia, Pennsylvania 19104
L. A. Jarvis and A. C. Hartford Division of Radiation Oncology, Department of Medicine, Dartmouth-Hitchcock Medical Center, Norris Cotton Cancer Center, Lebanon, New Hampshire 03756
(Received 4 May 2011; revised 19 May 2011; accepted for publication 19 May 2011; published 23 June 2011) Purpose: The authors hereby notify the Radiation Oncology community of a potentially lethal error due to improper implementation of linear units of measure in a treatment planning system. The authors report an incident in which a patient was nearly mistreated during a stereotactic radiotherapy procedure due to inappropriate reporting of stereotactic coordinates by the radiation therapy treatment planning system in units of centimeter rather than in millimeter. The authors suggest a method to detect such errors during treatment planning so they are caught and corrected prior to the patient positioning for treatment on the treatment machine. Methods: Using pretreatment imaging, the authors found that stereotactic coordinates are reported with improper linear units by a treatment planning system. The authors have implemented a redundant, independent method of stereotactic coordinate calculation. Results: Implementation of a double check of stereotactic coordinates via redundant, independent calculation is simple and accurate. Use of this technique will avoid any future error in stereotactic treatment coordinates due to improper linear units, transcription, or other similar errors. Conclusions: The authors recommend an independent double check of stereotactic treatment coordinates during the treatment planning process in order to avoid potential mistreatment of patients. C 2011 American Association of Physicists in Medicine. [DOI: 10.1118/1.3598444] V
I. INTRODUCTION Surgical and nonsurgical frame-based stereotactic radiosurgery (SRS) and fractionated stereotactic radiotherapy (SRT) systems with modified linear accelerators have been used for treatment of a wide variety of intracranial lesions since the late 1980s. Several stereotactic frame systems with specific localizers are commercially available and are modeled in a variety of radiation therapy planning systems. Typically, a localizer is fitted to the patient’s head frame prior to acquiring the treatment planning CT or MRI image, making it possible to precisely locate the individual image slices and the reference markers (rods) in the frame-specific coordinate space. Thus, the treatment target and critical structures shown in the CT or MRI images can be exactly delineated in the frame space using the treatment planning system. The localization algorithms6,7 are incorporated into the treatment planning systems, which transform the three dimensional coordinates of the treatment beam isocenter from treatment planning space to stereotactic frame space. The stereotactic 4108
Med. Phys. 38 (7), July 2011
frame is then used directly to position the patient’s head for treatment with high accuracy in the treatment room. We have used the Philips Pinnacle3 (Ref. 1) treatment planning system in combination with surgical and nonsurgical stereotactic head frames from Integra-Radionics.2 The BRW stereotactic localization system is incorporated into the head frame and is modeled in the planning system. The Pinnacle treatment planning system was originally designed for use in stereotactic radiosurgery, and this portion of the software platform has remained unchanged for many years. A review of the planning system was performed by the manufacturer as part of their software QA process and as of version 9.0, a change was made so that all linear coordinates are now reported in centimeters (cm). According to the release notes, this change was made to be compliant with IAEA technical report 1540.3 IAEA technical report 1540 states that linear coordinates must be reported in either millimeter (mm) or centimeter but not both in a planning system. However, stereotactic radiotherapy systems are specifically excluded from the requirements of the report. To quote from
0094-2405/2011/38(7)/4108/4/$30.00
C 2011 Am. Assoc. Phys. Med. V
4108
4109
Gladstone et al: Units of measure in stereotactic radiotherapy planning
that report: “Issues related to intensity modulated radiation therapy or other specialized techniques such as stereotactic radiosurgery are not addressed in this acceptance testing report.” The decision to implement this particular requirement in the Pinnacle planning system without regard to specific application resulted in stereotactic coordinates being reported by the system in units of centimeter rather than millimeter. A factor of 10 must, therefore, be applied to the printed treatment coordinates in order to be consistent with the stereotactic head frame, which has units of millimeter physically engraved in the frame hardware. II. METHODS The Pinnacle treatment planning system was upgraded from version 8.0 m to version 9.0 by the vendor via internet connection to our software server. Post upgrade testing was performed in accordance with TG-53 (Ref. 8) by reviewing baseline dose calculation accuracy and performing end to end testing of new features pertinent to our clinical operations as listed in the release notes “New features” section. The change in linear units from millimeter to centimeter was noted in the release notes in a section “Clarification for use;” however, this change was not noted or tested by the physicist performing acceptance testing. The Pinnacle treatment planning system was used to design an SRT plan for a patient suffering from a right acoustic neuroma. A relocatable Gill–Thomas–Cosman frame system4 was chosen for daily target localization and head refixation. The optimal treatment plan was approved by the physician and treatment setup sheets, including the BRW setup coordinates, were prepared by the dosimetrist for use by the therapists at the time of treatment. These coordinates were printed by the Pinnacle planning system in units of centimeter and were accurately transcribed as such by the dosimetrist into the setup notes. All paperwork was checked for accuracy and consistency by a physicist. At the time of treatment, the therapists set stereotactic localization equipment to the numbers shown on the setup sheet, not realizing that the units of measure had been recorded in centimeter rather than millimeter. Before the treatment was delivered, KV images were acquired to verify
and document the setup accuracy. On analysis of the setup images, it was found that the patient was misaligned in treatment space, and the therapeutic treatment was not delivered. A physicist was called to the treatment machine and the physicist’s investigation revealed the cause of the setup error. Retrospective analysis of this incident was performed to determine the potential severity had the error gone undetected and treatment been delivered. This analysis was performed by moving the treatment isocenter in the planning system to the improper coordinates, setting the monitor units to those of the original plan, and then calculating the dose which would have been delivered. The results of this calculation are shown graphically in Fig. 1 and in dose volume histograms, calculated for the acoustic neuroma and for the brainstem in Fig. 2. Undetected, this error would have inadequately treated the target and 5.7% (1.2 cc) of the brainstem would have received a dose of 50 Gy or higher. About 40% of the brainstem would have received a dose of 20 Gy or higher in the SRT. Clearly, such errors would damage the brainstem and would not adequately control the tumor. Procedures with such errors are intolerable and a robust method to verify the accuracy of stereotactic coordinates is needed before the patient is set up for treatment at the machine. Li et al.5 have reported a simple and accurate method to determine stereotactic coordinates based on digitization of the stereotactic localization system and the target point (isocenter) in the CT or MRI planning image sets. We have implemented their calculation formalism in a Visual Basic macro script to be run within Microsoft Excel. The user enters the planning system coordinates of each BRW localizer rod from two image slices and the coordinates of the treatment beam isocenter, then the macro script calculates the BRW coordinates to be used as verification against planning system determined coordinates. This image-based QA program allows us to determine the coordinate of any point of interest and detect such big errors or even a small uncertainty for the point localization in a planning system. This code is available upon request. In implementing Li’s algorithm, it was found that there are typographic errors in the original publication. The corrected formulas are as follows:
FIG. 1. Isodose distributions in the intended treatment (left) and to be delivered if the error had not been found (right). Medical Physics, Vol. 38, No. 7, July 2011
4109
4110
Gladstone et al: Units of measure in stereotactic radiotherapy planning
4110
FIG. 2. Dose volume histograms showing the intended dose (solid lines) and the dose which would have been delivered had the error not been caught (dashed lines) to the target (blue) and the brain stem (purple). This error could yield damage of 40% of the brainstem with dose of >20 Gy while leaving the entire target untreated with its maximum dose of only 10 Gy.
From Eq. (1):
Y8 ¼ D cos 30 From Eq. (6):
b1 ¼ ½x2 x8 ðX2 X5 Þ ðx2 x5ÞðX2 X8 Þ=T b2 ¼ ½x2 x8 ðY2 Y5 Þ ðx2 x5ÞðY2 Y8 Þ=T b3 ¼ ½x2 x8 ðZ2 Z5 Þ ðx2 x5ÞðZ2 Z8 Þ=T Please see the original publication for a detailed description of these formulas. The output of this implementation of Li’s algorithm was compared to that of the BRW coordinates calculated by the Pinnacle planning systems (converted to millimeter) for ten clinical cases to verify its accuracy. Each CT was scanned with 2.5 mm slice resolution. III. RESULTS Table I contains the BRW coordinates in the three orthogonal directions for the ten cases studied. The agreement between the Pinnacle algorithm and our implementation was found to be within 0.42 mm in the AP direction, 0.54 mm in the lateral direction and 0.65 mm in the Vertical direction with standard deviations of 0.17 mm, 0.26 mm, and 0.48 mm, respectively, thereby verifying the accuracy of this implementation. IV. DISCUSSION The clinical error and near miss reported in this letter are a direct result of misapplication of IAEA requirements for planning system units of measure to stereotactic radiotherapy systems. Testing of new features in the software version upgrade did not uncover the change in software pertaining to Medical Physics, Vol. 38, No. 7, July 2011
units of measure and inadequate attention was given to the “Clarification for use” section of the release notes. This misapplication of regulations went undetected until the time of treatment because: 1. The change in units of measure for SRS coordinates was not detected during post software update testing as described above. Inadequate attention was given to the Clarifications for use section of the software release notes. 2. Notice of the change in software was provided by the manufacturer on page 23 of the release notes in a section labeled “Clarifications for use” rather than prominently highlighted in the section “New features” at the beginning of the document. Additionally, the release notes were supplied as a PDF file in the system documentation disk, rather than in a booklet as has been past practice. TABLE I. BRW coordinates calculated by the macro reported in this report, a commercial planning system and the agreement. All numbers are in units of millimeter. This report Case 1 2 3 4 5 6 7 8 9 10
AP 54.6 5.6 21.3 4.2 6.0 31.7 12.2 9.3 5.8 24.7
LAT VERT 7.8 10.2 33.6 31.2 21.2 38.2 12.3 18.0 10.5 49.0
Commercial AP
LAT VERT
Delta AP
LAT VERT
15.9 54.7 7.7 15.6 0.1 0.1 64.4 5.7 10.3 64.8 0.1 0.1 2.2 21.5 33.1 1.6 0.2 0.5 15.2 4.6 31.6 14.6 0.4 0.4 26.7 6.0 21.0 26.0 0.0 0.2 25.2 31.8 38.1 25.3 0.1 0.1 11.8 12.2 12.5 11.3 0.1 0.2 31.1 9.3 17.8 30.5 0.0 0.2 65.3 5.7 10.3 64.8 0.1 0.2 38.4 24.7 48.9 38.0 0.0 0.1
0.3 0.4 0.6 0.5 0.6 0.1 0.5 0.6 0.5 0.4
4111
Gladstone et al: Units of measure in stereotactic radiotherapy planning
3. The planning system upgrade was performed by the vendor via remote system access. No vendor representative was on site for the purposes of acceptance testing of the software, no acceptance testing document was provided by the vendor, and no review of software compliance was made as required by IAEA 1540. 4. Units of centimeter are commonly used in the clinical environment and their use on the setup sheet was not detected as abnormal by any staff member. 5. The treatment therapists are unaccustomed to checking units of measure and should not be expected to do so as values should be provided to them in units suitable for direct use. 6. The head frame system is obviously inscribed in units of millimeter, however, the actual unit millimeter is not inscribed on the system, only the marks and numerical values are inscribed. Attention is therefore not drawn to the actual unit. 7. Careful focus is required to read the vernier system on the head frame localizer, attention is naturally given to this fine detail without thought to the units of measure.
of each item in software release notes by initials of the reviewing physicist. This review is in addition to our normal TG53 preclinical testing. As a final control, we have removed version 9.0 of the Pinnacle software from default clinical use until such time that a software fix is available.
V. CONCLUSION We have identified a potentially lethal error in a radiation therapy planning system with the root cause being use of units of linear measure, which are inconsistent and inappropriate for the physical systems being modeled. We have implemented a method of redundant, independent checking of stereotactic coordinates which can detect such an error or similar errors in the future. We have altered our software testing and acceptance policies to extend reviews beyond those explicitly listed in TG53. Finally, we have identified and corrected a typographical error in the literature with respect to the mathematical equations used in these solutions. a)
Because of the points listed above, this error is insidious and can easily go undetected in the clinical setting. Had pretreatment imaging not been used in the case reported, the error would have reached the patient and could have resulted in serious harm or death. We, therefore, have instituted a policy of independently checking stereotactic coordinates before committing them to the patient setup records. When possible, the use of pretreatment imaging (IGRT) to verify proper patient setup is strongly recommended as a last chance to avoid large geometric treatment errors. We urge other users to consider similar procedures to avoid improper determination of such coordinates, transcription, or other similar errors. Notwithstanding IAEA 1540, we implore treatment planning vendors to model and report all treatment coordinates in units which are consistent with those of the hardware being modeled. Redundant quality assurance steps (IGRT) prevented harm to any patient in this case. Clearly, however, our preclinical testing procedure for software upgrades was inadequate to detect this error. We have, therefore, implemented a policy of line item review and explicit acknowledgement
Medical Physics, Vol. 38, No. 7, July 2011
4111
Author to whom correspondence should be addressed. Electronic mail:
[email protected] 1 Philips Medical Systems, 5520 Nobel Drive, Suite 125 Fitchburg, Wisconsin. 2 Integra-Radionics, 311 Enterprise Drive, Plainsboro, New Jersey 08538. 3 IAEA TECH-DOC 1540, “Specification and acceptance testing of radiotherapy treatment planing systems,” April 2007. 4 H. M. Kooy, S. F. Dunbar, N. J. Tarbell, E. Mannarino, N. Ferarro, S. Shusterman, M. Bellerive, L. Finn, C. V. McDonough, and J. S. Loeffler, “Adaption and verification of the relocatable Gill-Thomas-Cosman frame in stereotactic radiotherapy,” Int. J. Radiat. Oncol. Biol. Phys. 30(3), 685– 691 (1994). 5 S. Li, J. F. Jackson, L. T. Myers, N. A. Detorie, and J. F. Dicello, “A simple and accurate coordinate transformation for a stereotactic radiotherapy system,” Med. Phys. 26(4), 518–523 (1999). 6 C. B. Saw, K. Ayyangar, and N. Suntharalingam, “Coordinate transformations and calculation of the angular and depth parameters for a stereotactic system,” Med. Phys. 14(6), 1042–1044 (1987). 7 P. Grunet, J. Maurer, and W. Muller-Forell, “Accuracy of stereotactic coordinate transformation using a localisation frame and computed tomographic imaging. Part 1. Influence of the mathematical and physical properties of the CT on the image of the rods of the localisation frame and the determination of their centers,” Neurosurg. Rev. 22, 173–187 (1999). 8 B. Frass, K. Doppke, M. Hunt, G. Kutcher, G. Starkschall, R. Stern, and J. Van Dyke, “American association of physicists in medicine radiation therapy committee task group 53: Quality assurance for clinical radiotherapy treatment planning,” Med. Phys. 25(10), 1773–1829 (1998).
