A Computer-Graphic Display for Real-Time Operator Feedback during Interventional X-Ray Procedures

Kevin Chugh1, Petru Dinu1, Daniel R. Bednarek1, Darold Wobschall2, Stephen Rudin1, Kenneth Hoffmann1, Ron Peterson2, Ming Zeng2 1 State University of New York at Buffalo, Amherst, New York 14260 2 Esensors, Inc., P.O. Box 1702, 4240 Ridge Lea Rd. Amherst, NY 14226
ABSTRACT
The harmful effects of ionizing radiation, as employed in a variety of medical imaging procedures, have been well studied and documented. To minimize risk to patients, operators must continually assess the dose rate and cumulative dose to the patient at each area of exposure. We have developed a computer graphic dose management display system which provides this operator feedback. The system is comprised of a signal processing module which reads the state of a fluoroscopy machine, a transmission ionization chamber for exposure measurement, and a visualization of the patient that displays the current level of radiation intensity and accumulated dose at every location on the body. The system shows the beam projection and orientation of the machine and color-coded dose metrics on the patient graphic model in real time. Additionally, a database system has been incorporated to allow for recording and playback of the entire procedure.

1. INTRODUCTION
The deleterious effects of ionizing radiation are well documented and include both stochastic and deterministic effects1. In the past, the primary concern during diagnostic x-ray procedures has been for stochastic long-term effects such as carcinogenesis. Recently, there has been increasing concern for the short-term acute effects resultant from high doses of radiation associated with long interventional procedures which use fluoroscopic guidance. These acute effects are deterministic and generally have a threshold below which the effect is not observed. Effects which have been documented include epilation and serious skin injuries such as moist desquamation and tissue necrosis2. To make the medical community aware of this potential for severe acute effects, the FDA has issued several advisories3, 4which recommend the determination of radiation dose, explaining that the patients medical record should contain an unambiguous identification of those areas of the patients skin that received an absorbed dose that may approach or exceed the selected threshold. To monitor patient dose, several types of dosimeter have been used. Large-area, transmission ionization chambers can be used to measure the product of dose to air and x-ray beam area or the dose-area-product (DAP). Such chambers have a sensitive cross-sectional area which is larger than the x-ray field and, when placed between the x-ray source and the patient, can indicate the total quantity of radiation entering the patient. This quantity has been shown to be directly related to energy imparted5-7 and can be a good indicator of risk of stochastic effects8, 9. We have built such chambers and previously reported on their use to monitor fluoroscopic procedures10-13. However, these chambers do not measure skin dose and cannot provide an indicator of deterministic risk. Several methods have been proposed for directly measuring skin dose including the use of thermoluminescent dosimeters (TLDs) and x-ray fluorescent scintillators with a fiber optic light guide (Model TN-RD 50, Thomson & Nielson Electronics, Ltd., Ottawa, Canada). These dosimeters only give dose values at the individual points where they are located on the skin and require the detectors to be placed in the radiation field where they may be observed in the image and be distracting to the interventionist. Two transmission ionization chambers which can be used to measure radiation dose14, 15 have been

proposed and are similar to that which has been built by us11. With a fixed source to patient skin distance (SSD), these types of dosimeters can provide a measurement of skin dose. However, in a dynamic fluoroscopic interventional procedure, SSD changes as well as the location on the patients skin that is being exposed. A measurement of cumulative radiation concentration at the output of the x-ray tube (as these chambers measure) is not sufficient to monitor actual skin dose. It is the goal of the system described here to measure the output intensity of the x-ray beam and to track the location on the patient which is absorbing the radiation energy. In this way skin dose can be tracked while the beam is moved over the patient. In many procedures, the x-ray beam remains fixed over a given location of the patient for a considerable period of time while the interventional action is occurring. In this case, tracking the location on the skin which is exposed during the procedure can be used as a tool to achieve skin dose reduction. Skin dose can be reduced by deliberately moving the x-ray beam over the skin surface during the procedure so that the entrance point is changed while the patient feature of interest remains in the field of view of the image receptor. We have referred to this technique as dose spreading. 16 For dose spreading to be successful, the dose needs to be tracked and correlated to location on the entrance skin surface as a graphical (computer) display. By providing the operator with the dose history over the patient skin, he or she can be guided as to the optimal beam direction for maintaining the skin dose below threshold values for deterministic effects.

2. TECHNICAL APPROACH
2.1 Architecture The dose tracking system as described previously17 has been designed with 3 main components- a dosimeter, which reads the physical dose in proximity to the x-ray tube, a digital signal processor, which reads the state of the fluoroscopy machine, and a software application that fuses these data streams and gives a visual display of the current dose rate and accumulated dose to the patient. Figure 1 illustrates the architecture of the system.

Figure 1. System architecture.

2.1.1 Fluoroscopy Machine The dose tracking system is being developed using a Toshiba CAS 8000 V machine, which has an analog interface for tapping the states of the machine. The output from this analog tap is fed into an analog to digital converter and read into the computer processing the signals with a USB interface. In the prototype system, this is the same computer that is running the graphic software application itself, but the architecture is flexible and modular, so this need not be so. The data being read from the analog tap is a voltage signal that needs further calibration to obtain the desired values. This is discussed later in the software section. The information read from the fluoroscopy machine includes the following: the current values of each positional degree of freedom of the C-Arm and table, the current and voltage of the x-ray tube, and the NTSC video feed of the x-ray image. We have developed this system for a 9 degree of freedom system (the Toshiba CAS 8000 V) - the gantry has 3 compound rotational degrees of freedom and 2 translation degrees of freedom and the table has 1 rotational degree of freedom and 3 translational degrees of freedom. The system is readily adaptable to virtually any mechanical configuration. These degrees of freedom are illustrated in Figure 2.

Figure 2. Mechanical degrees of freedom of the gantry and patient table. In addition, the x-ray tube current and electric potential difference is read so that the x-ray output can be indirectly monitored. These values are measured from signals on the machines analog signal tap. These readings are intended to be redundant to the dosimeter and are used for verification and calibration, but may also be used as an alternate means of x-ray output tracking with proper calibration. Finally, the analog video output is tapped for use in the beam shape determination. Figure 4 shows a sample output captured for use in the shape determination software module. A separate binary output is available to determine the image intensifier magnification mode and the source-to-image distance is read out using an analog tap which is A/D converted so that the beam dimensions can be determined.

2.1.2 Dosimeter The dosimeter used is a transmission ionization chamber which can measure the integral output (dose area product) and the beam intensity (dose) 11and is placed on the outside of the x-ray tube collimator. Beam area can be determined from the ratio of the two chamber outputs. A custombuilt dual electrometer and microcontroller reads out the current and transmits the exposure data to the PC in digital format. 2.1.3 Dose Tracking Application The dose tracking application reads the input data, performs computations, and gives visual feedback to the operator and records the procedure for historical record. Within the software application itself, there are 6 main components- signal processing, beam shape determination, computational engine, patient model, database module, and graphic display. These modules and their interactions are illustrated in Figure 3. Each of these modules is described in detail in the following section. In Figure 3, these modules are represented by rectangles, while data sources are trapezoids.

Figure 3. Components of the software application.

2.1.3.1 Signal Processing The signal processing module reads the current state of the machine and filters the data stream. The filtering is comprised of accumulating and averaging in order to reduce noise in the data. The degrees of freedom shown in Figure 2, as well as the current exposure state of the x-ray tube itself are then sent to the computational engine. Additionally, the signal processor module reads the dosimeter values and, coupled with the x-ray state reported from the fluoroscopy machine directly computes a calibrated entrance skin dose for single spot exposures or dose rate for fluoroscopic exposures for each point of the patient in the beam. 2.1.3.2 Beam Shape Determination In order to accurately reflect the geography of the physical exposure in a computer model, x-ray beam shape information is necessary. Since the machine that this system is interfaced with does not track the collimator position, we have devised a method to determine the beam shape from the x-ray image. The beam-shape determination module reads the NTSC video output of the fluoroscopy machine and using an edge detection technique, computes the position of the beam edges, and their orientation as shown in Figure 4. This information is also fed to the computational engine.

Figure 4. The raw video-captured image and the final shape determination computed by the edge detection module. 2.1.3.3 Database Module The database module acts as a data broker between the state of the exam (patient, fluoroscopy machine, accumulated exposure by geography), and the database. The exam begins when an operator enters the patient's personal information. The operator then signals that the software should begin recording the procedure by clicking a button. The database module then continually queries the computational engine and the patient model for changes in state and records those changes along with the current system time. This data capture is then periodically committed to the database. When an exam is being reviewed, the reverse process occurs- the database module reads the state entries from the database tables and periodically (as a function of the recorded timestamp of each state change) sends new states to the patient model and computational engine, which in turns update the display to reflect those changes.

2.1.3.4 Computational Engine The computational engine, as can be seen from Figure 3, reads input from the signal processing module, and the beam shape determination module and computes 3 things- the orientation in world space of the gantry, and thus the x- ray source, the "dose volume" which describes the volumetric loci of the beam to be tested for intersection with the patient, and the exposure intensity at the x-ray source. The transformation of the gantry is determined by the following equation:

Tg

Tb RGy RGx RGzTT 1RTz1

where Tb represents the x-ray focal spot height above the origin referenced to the center of the table,

RGi represents the rotation of the gantry about the i axis (X, Y, or Z), TT 1 is the inverse compound
translation matrix composed of the product of the three table translational degrees of freedom and RTz is the inverse table rotational matrix about Z. The dose volume is simply the set of vectors from the x-ray focal spot in world coordinates to each point of the projected beam shape. 2.1.3.5 Patient Model The patient model module accomplishes 3 goals. First, it provides a representation of the geographic accumulation of exposure on the skin. The computational engine continually sends the patient model a data stream consisting of the current radiation state of the machine, and the current "dose volume" determined by the mechanical state of the machine and the beam shape. With this data, the patient model module determines which areas on the skin surface are exposed, and accumulates dose at these surface elements. The patient-model module first determines which vertices are affected by testing for collision between the dose volume and the patient geometry. Each vertex's accumulated skin dose is then computed using a calibrated measured exposure value at a reference point, inverse square correction to the position of the skin, a backscatter factor which is a function of field size and kVp, and the appropriate dose conversion factor. Secondly, the patient model module interfaces with the database module to keep a record of the exam for future review. The patient's personal information, along with the state changes of the machine, and the geographic exposure are all recorded. The data can be retrieved readily for playback or analysis. The final function of this module is to send the patient geometry and dose state of each vertex of the patient model to the graphic display module for visual rendering, as shown in Figure 5.
1

Figure 5. Visual representation of a patient with the highlighted area (see arrow) showing the location where the patient has been exposed to radiation. A color coded shading indicates the level of either cumulative dose or current dose rate during fluoroscopy. 2.1.3.6 Graphic Display Each of the modules described in this section ultimately contributes to the graphic display, which computes the visual representation of the state of patient and the fluoroscopy machine. This module coordinates the "drawing" of the patient and fluoroscopy machine and paints the resulting image onto the computer display. Essentially, this module is an encapsulation of a graphics state machine, and employs OpenGL18, 19 to render the image. This module also processes user input as it relates to the display, allowing the operator to zoom and rotate the rendering of the patient and fluoroscopy machine

3. RESULTS
The dose tracking system in its current state can accurately reflect the geographic specific exposure of a patient during an interventional x-ray procedure. Figure 6 shows a snapshot of the software application during one of these procedures.

Figure 6. Example of display during procedure showing current beam projection on patient, dose levels and system positional coordinates. Not only is a visible color coded representation of the patient's skin dose displayed to the operator, certain metrics, such as average dose and maximum dose at the currently beam directed area, are also displayed. The patient graphic can be zoomed in or out to show the desired detail in the localized area where the beam intersects the patient. In addition, patient information and feedback about the current mechanical state of the machine can also be displayed on a split screen as shown in Fig. 6. The patient graphic can optionally cover the entire screen for greater detail. Also, a procedure in its entirety including beam position and dose history can be captured and played back for review. The procedure playback is virtually indistinguishable from the original procedure in the virtual representation environment and provides a visually and physiologically accurate representation of the progression of a procedure. 4. CONCLUSIONS AND FUTURE WORK 4.1 Summary We have developed a real-time, visually-represented dose tracking system that provides operator feedback during an interventional x-ray procedure. This is accomplished by displaying a patient model with a color coded indication of accumulated dose, along with metrics that display the implications of additional

exposure. The system is comprised of real time data acquisition, filtering, computation, and graphic display. All data is recorded for post procedure review and visual playback. 4.2 Future work There are several enhancements currently in development or planned for the dose tracking system. The main ones are discussed below. 4.2.1 Anthropometry There is a fairly well-studied and well-documented variation in the physical shape of human beings20, 21, and this should be reflected in any virtual representation of a patient. Work is currently being done to implement an anthropometry model that is accurate to an acceptable level of error, but at the same time parameterized as to require minimal operator input. This model will be incorporated into the dose tracking system and will dramatically improve not only visual accuracy, but also geographic correspondence between the virtual model and that patient. 4.2.2 Articulated Model Along with improving the fidelity of the human representation, the mechanical capabilities of the virtual representation must be improved. This will be done by implemented a multi-jointed articulated model so that a patients position can be more accurately represented. This will improve the general accuracy of the representation, and will also allow for procedure specific body configurations, such as a patient holding an appendage of their body in a particular position during the procedure. 4.2.3 Direct Digital Interface A digital output interface containing information on the state of the fluoroscopy machine is available on current Toshiba C-Arm units. This will allow us to obtain the information we are now obtaining though the analog tap in direct digital form and will simplify the data acquisition. We are currently working on this adaptation. ACKNOWLEDGEMENTS This work was supported by FDA grant # R44 FD-01584-02 and Esensors, Inc. and an equipment grant from Toshiba Medical Systems, Inc.

REFERENCES
1. 2. 3. 4. 5. Wagner, L., E. PJ, and G. RA, Potential Biological Effects Following High X-ray Dose Interventional Procedures. Journal of Vascular and Interventional Radiology, 1994. 5: p. 71-84. Shope, T., Radiation Induced Skin Injuries from Fluoroscopy. Radiology, 1996. 197(Supplement): p. 449. FDA, Avoidance of Serious X-ray Induced Skin Injuries to Patients During FluoroscopicallyGuided Procedures. US Food and Drug Administration, 1994. FDA, FDA Public Health Advisory: Avoidance of serious x-ray induced skin injuries to patients during fluoroscopically-guided procedures. US Food and Drug Administration, 1994. Alm Carlsson, G. and C. Carlsson, Relations Between Effective Dose Equivalent and Mean Absorbed Dose (Energy Imparted) to Patients in Diagnostic Radiology. Physics in Medicine and Biology. 31. Carlsson, C., Determination of Integral Absorbed Dose From Exposure Measurements. Acta Radiologica Therapy, Physics, Biology, 1963: p. 433-58. Pychau, P. and E. Bunde, The Absorption of X-rays in a Body Equivalent Phantom. British Journal of Radiology, 1965. 38: p. 875-877. Le Heron, J., Estimation of Effective Dose to the Patient During Medical X-ray Examinations from Measurements of the Dose-Area Product. Physics in Medicine and Biology, 1992. 37: p. 21172126. Hart, D. and B.F. Wall, Estimation of Effective Dose from Dose-Area-Product Measurements for Barium Meals and Barium Enemas. British Journal of Radiology, 1994. 67: p. 485-489. Bednarek, D. and S. Rudin, Comparison of Two Dose-Area-Product (DAP) Ionization Chambers. Medical Physics, 1997. 24: p. 1356. Bednarek, D. and S. Rudin, A Double Transmission Ionization Chamber for Dose and Dose-AreaProduct Monitoring. Medical Physics, 1998. 25(8): p. A1585. Bednarek, D., S. Rudin, and R. Wong, Assessment of Patient Exposure for Barium Enema Examinations. Invest Radiol, 1983. 18: p. 453-458. Bednarek, D., et al., Reduction of Fluoroscopic Exposure for the Air-Contrast Barium Enema. British Journal of Radiology, 1983. 56: p. 823-828. Mahesh, M., T. Beck, and N. Azar, Radiation Dose Monitor for Fluoroscopic Procedures. Medical Physics, 1997. 24(6): p. 991-992. Gfirtner, H., F. Stieve, and J. Wild, A New Diamentor for Measuring Kerma-Area Product and Air Kerma Simultaneously. Medical Physics, 1997. 24(12): p. 1954-59. Rudin, S. and D. Bednarek, Comparison of Filter Material and Design for use in ROI Angiography. SPIE Physics of Medical Imaging, 1993. 1896: p. 354-364. Dinu, P., et al., X-Ray Beam Tracking System for a Fluoroscopic C-Arm Unit. Medical Physics, 2003. 30(6). Angel, E., OpenGL : A Primer. 2002, Boston ; Toronto: Addison-Wesley. Woo, M., et al., OpenGL Programming Guide : The Official Guide to Learning OpenGL, version 1.1. 2nd ed. 1997, Reading, Mass. Don Mills, Ont.: Addison Wesley. xxxviii, 650 p. Montagu, A., A Handbook of Anthropometry. 1960, Springfield, Ill.,: Thomas. 186. Easterby, R.S., K.H.E. Kroemer, and D.B. Chaffin, Anthropometry and Biomechanics : Theory and Applications. NATO conference series. III, Human factors. v. 16. 1982, New York: Plenum Press. x, 327.

6. 7. 8.

9. 10. 11. 12. 13. 14. 15. 16. 17. 18. 19. 20. 21.