Inr. J. Radiation Oncology Biol. Phvs. Vol. 4. pp. 1107-l Printed in the U.S.A. @ Pergamon Press Inc.. 19%
109
??Technical Innovation and Note A FEASIBILITY STUDY OF COMPUTERIZED TOMOGRAPHY RADIATION THERAPY TREATMENT PLANNING IN TRANSVERSE, CORONAL AND SAGGITAL SECTIONS RUSHDY ABADIR, M.D. and F. MARC EDWARDS, Ph.D. Department of Radiology, Section of Radiation Oncology, University of Missouri Medical Center
and GREGORY Bioengineering/Advanced
LARSEN,
Ph.D.
Automation, University of Missouri, Columbia. MO 65212, U.S.A.
The feasibifity of simultaneous coronal or sag&al reconstruction of computerized axial tomography (CT) images and isodose dkfbutioas from multLptanar data has been examined as an approach to three-dimensional treatment pianning. With the aid of a computer (PDP11/50), serial se&ens of a CT were wed to reconstruct saggitpl and coronal secthms. Ceutral and off-center fsodose distributions on tbe conqmmding cross section were produced on a radhtioa therapy treatment planning (Artronix PC-12) computer. Isodose curves thus produced were used for reconstn~~&n of mggital and coronal isodose distributions. Tbe limitations of tbat system are dffussed as well as further developments.
INTROIIUCTION The exact spatial distribution of tumors, both in single planar and three-dimensional sense, has been of great concern to the radiation therapist. The desire to avoid “geographic miss” may suggest using large fields, while the necessity to preserve normal tissues may dictate small fields. The ultimate plan of treatment usually is a balance between the two factors. When computerized axial tomography (CT) was introduced, it was seen as a potential means of identifying the exact extent of the tumor in cross sections at various levels. A further application of multiple level CT scans is to use this information for computerized treatment planning to produce three-dimensional distributions both of the tumor and the isodose curve. Many authors’-3 have described CT-radiation therapy systems of various degrees of sophistication. This paper presents the result of a trial of the feasibility of saggital and coronal section reconstruction of the CT images as well as the isodose distribution.
the tumor starting from the base of the skull upwards. Each section was 8 mm thick with 2 mm table increment between scans. Thus 40 CT scans were necessary for 8 cm of field length. This procedure required several hours of scanning time. The resulting scans were transferred to magnetic tape for further processing. An isodose distribution was calculated for each section in the following manner. First the scans were displayed off-line using a computer (PDP 1l/50) with a fixed head, black and white disc display TV system. Photographs of each section were enlarged to life size by determining the correct scaling factor from known water bath head cone dimensions. The enlarged sections were checked for accuracy by comparing the resulting bi-parietal diameter with the true diameter as measured from the patient. The resulting photographs were used as patient contour input data for the external beam program of a radiation therapy planning (Artronix PC-12) computer. In order for true saggital and coronal reconstruction to occur, the transverse sections had to be in the same relative orientation: thus, an identical coordinate system was used for all sections. The tracing device (PC-12 thetaphi graphics input) coordinate system was defined for
METHODS AND MATERIALS A CT(EMI) head scanner was used to obtain multiple overlapped scans of the patient’s head in the region of Presented at the American Society of Therapeutic Radiologists Meeting in Denver, Colorado. Works in Pro-
gress, November 1977. Accepted for publication 4 August 1978. 1107
1108
Radiation Oncology 0 Biology 0 Physics
each contour by requiring the water bath head cone image, which appeared as a ring around the skull, to be identically oriented for all levels. If patient motion occurred between scans, a slight change in coordinate systems was required to maintain anatomical fidelity. Hence, when necessary, the relative position of sequential sections was related to anatomical landmarks common to both scans. Once all components were entered in the radiation therapy computer, preliminary treatment planning was performed in the plane containing the central axis. Off-axis beams were generated for each off-axis patient contour and two-dimensional isodose distributions were computed for each section. Hence this required that the radiation beams be located in identical relative positions for all sections; this task was accomplished manually by marking the beam entry points on all patient contours. The plane of isodose calculation was required to be coplanar with the CT scan. Thus the radiation beam orientation was variable only in plane of the CT scan and the central ray could not be positioned at oblique angles with respect to the transverse image planes. Isodose lines were calculated at 10% intervals and subsequently plotted at full scale by incremental plotter. Finally, the isodose distribution for each CT section was entered in PDP 11/50 computer by tracing each isodose line with a graf-pen. The computer used the image scaling factors and coordinate systems as previously described to combine the CT scan with the isodose distribution by assigning an isodose value to the locus of pixels which were most spatially coincident with each isodose line. RESULTS A representative transverse scan with the isodose distribution resulting from a 45” wedge pair is shown in Fig. 1. Saggital (Fig. 2) and coronal (Fig. 3) views were reconstructed by requesting the computer to
November-December 1978. Vol. 4, No. 11 and
NO. 12
Fig. 2. Computer reconstructed saggital section of image and isodose curves.
Fig. 3. Computer reconstructed coronal section of image and isodose curves.
retrieve from each transverse section, the appropriate pixels which make up the desired plane. The computer displayed the CT scan in monochrome grey-scale while the isodose lines could be displayed either in polychrome or monochrome grey-scale (using a color different from that of CT scan by assigning a different color or shade to each isodose value). DISCUSSION One of the first uses of CT information for radiation therapy treatment planning was reported by Chernak et al.,’ who utilized CT scans for planning in single transverse planes. Several other authors’” have published similar results with the addition of greater utilization of CT data for inhomogeneity corrections. Initial results of the clinical efficacy of CT use in radiotherapy have been reported by Munzenrider et ~1.~In a recent report of the Committee on Radiation Oncology Studies Sub-committee on CT scanning and
Fig. 1. Computer display of treatment plan on a cross section at the level of the central beam.
Radiation Therapy6 sional calculation
the importance and display
of three-dimentechniques was
Study of computerized tomography radiation therapy treatment ?? R. AB~DIRet al.
emphasized. While true three dimensional radiation beam models do exist”’ the emphasis of most current treatment planning systems is on single or multiplanar calculation. Hence one approach, the one examined here, is to utilize multi-planar image and isodose information to generate coronal and saggital information. There are several problems with this demonstration system. A large number of scans were used. As shown previously,4 this close spacing was necessary in order to maintain a resolution in the reconstructed coronal and saggital planes which is equivalent to that in the transverse planes. The large number of scans is time-consuming and gives rise to motion and sequential orientation problems. These difficulties can be minimized by reducing the required number of scans and the time per scan. A fast CT scanner is required for a clinically feasible CT-RT system. In order to reduce the number of scans in the axial direction it is necessary to sacrifice spatial resolution of the CT image display of reconstructed planes and develop three dimensional treatment planning systems which can interpolate in all planar views. As previously noted, external beam treatment planning programs of most present radiotherapy computers interpolate only in single planar sections although 3D calculations are possible using an irregular field approach. True three-dimensional beam models and isodose interpolation based on a 1 cm cubic grid point
1109
system in conjunction with fast CT scanners probably would reduce the total scan time to approximately 2 min for 10 cm in the axial direction. The present system is limited to coronal and saggital planes which are othogonal to the transverse scan planes. Furthermore, it is feasible to position the beams only in planes parallel or perpendicular to the plane of transverse scans. A system that is clinically feasible should be able to position beams arbitrarily, to calculate isodose distributions, and to reconstruct pictures in any plane. Again this will require flexible three dimensional treatment planning software. The present system depends far too much upon operator action to define coordinate systems, orient scans, relative to each other and input data. The total time for a complete study, even after the CT scans are acquired, is at least 4 hr. This is not of concern for a feasibility study of this type; however it is unacceptable for a clinical system. Hence, for ease and speed of operation, it would be most convenient to have the image processing and treatment planning capability implemented on a single computer. This would allow a unified, simpler, approach to the solution of problems such as scaling, orientation of sequential sections, beam positioning, isodose calculations and display of the results. It would require a computer of greater capacity and speed than the one that presently is used exclusively for radiation therapy planning.
REFERENCES 1. Chemak, E.S., et al.: The use of computer tomography
for radiation therapy treatment planning. Radiology 117: 613-614, 1975. 2. Fullerton, G.D., et al.: CT determination of parameters for inhomogeneity corrections in radiation therapy of the esophagus. Radiology 126: 167-171, 1978. 3. Geise, R.A., McCullough, E.C.: The use of CT scanners in megavoltage photo-beam therapy planning. Radiology 124: 133-141, 1977. 4. Glenn, W.V. et al.: Image generation and display technique for CT scan data. Invest. Rad. 10: 403-446, 1975.
5. Munzenrider, J.E. et al.: Use of body scanner in radiology treatment planning. Cancer 40: 170-179, 1977. 6. Steward, J.R., Hicks, J.A., Boone, M.L.M., Simpson, L.D.: Computed tomography in radiation therapy. Int. J. Radiat. Oncol. Biol. Phys. 4: 313-324, 1978. 7. van de Geijn, J.: A computer program for 3-D planning in external beam radiation therapy, EXTDOS. Camp. Prog. Biomed. 1: 47-57, 1970. 8. Weinkam, J., Sterling, T.: A versatile system for the three-dimensional radiation dose computation and display, RTP. Comp. Prog. Biomed. 2: 171192, 1972.
109
??Technical Innovation and Note A FEASIBILITY STUDY OF COMPUTERIZED TOMOGRAPHY RADIATION THERAPY TREATMENT PLANNING IN TRANSVERSE, CORONAL AND SAGGITAL SECTIONS RUSHDY ABADIR, M.D. and F. MARC EDWARDS, Ph.D. Department of Radiology, Section of Radiation Oncology, University of Missouri Medical Center
and GREGORY Bioengineering/Advanced
LARSEN,
Ph.D.
Automation, University of Missouri, Columbia. MO 65212, U.S.A.
The feasibifity of simultaneous coronal or sag&al reconstruction of computerized axial tomography (CT) images and isodose dkfbutioas from multLptanar data has been examined as an approach to three-dimensional treatment pianning. With the aid of a computer (PDP11/50), serial se&ens of a CT were wed to reconstruct saggitpl and coronal secthms. Ceutral and off-center fsodose distributions on tbe conqmmding cross section were produced on a radhtioa therapy treatment planning (Artronix PC-12) computer. Isodose curves thus produced were used for reconstn~~&n of mggital and coronal isodose distributions. Tbe limitations of tbat system are dffussed as well as further developments.
INTROIIUCTION The exact spatial distribution of tumors, both in single planar and three-dimensional sense, has been of great concern to the radiation therapist. The desire to avoid “geographic miss” may suggest using large fields, while the necessity to preserve normal tissues may dictate small fields. The ultimate plan of treatment usually is a balance between the two factors. When computerized axial tomography (CT) was introduced, it was seen as a potential means of identifying the exact extent of the tumor in cross sections at various levels. A further application of multiple level CT scans is to use this information for computerized treatment planning to produce three-dimensional distributions both of the tumor and the isodose curve. Many authors’-3 have described CT-radiation therapy systems of various degrees of sophistication. This paper presents the result of a trial of the feasibility of saggital and coronal section reconstruction of the CT images as well as the isodose distribution.
the tumor starting from the base of the skull upwards. Each section was 8 mm thick with 2 mm table increment between scans. Thus 40 CT scans were necessary for 8 cm of field length. This procedure required several hours of scanning time. The resulting scans were transferred to magnetic tape for further processing. An isodose distribution was calculated for each section in the following manner. First the scans were displayed off-line using a computer (PDP 1l/50) with a fixed head, black and white disc display TV system. Photographs of each section were enlarged to life size by determining the correct scaling factor from known water bath head cone dimensions. The enlarged sections were checked for accuracy by comparing the resulting bi-parietal diameter with the true diameter as measured from the patient. The resulting photographs were used as patient contour input data for the external beam program of a radiation therapy planning (Artronix PC-12) computer. In order for true saggital and coronal reconstruction to occur, the transverse sections had to be in the same relative orientation: thus, an identical coordinate system was used for all sections. The tracing device (PC-12 thetaphi graphics input) coordinate system was defined for
METHODS AND MATERIALS A CT(EMI) head scanner was used to obtain multiple overlapped scans of the patient’s head in the region of Presented at the American Society of Therapeutic Radiologists Meeting in Denver, Colorado. Works in Pro-
gress, November 1977. Accepted for publication 4 August 1978. 1107
1108
Radiation Oncology 0 Biology 0 Physics
each contour by requiring the water bath head cone image, which appeared as a ring around the skull, to be identically oriented for all levels. If patient motion occurred between scans, a slight change in coordinate systems was required to maintain anatomical fidelity. Hence, when necessary, the relative position of sequential sections was related to anatomical landmarks common to both scans. Once all components were entered in the radiation therapy computer, preliminary treatment planning was performed in the plane containing the central axis. Off-axis beams were generated for each off-axis patient contour and two-dimensional isodose distributions were computed for each section. Hence this required that the radiation beams be located in identical relative positions for all sections; this task was accomplished manually by marking the beam entry points on all patient contours. The plane of isodose calculation was required to be coplanar with the CT scan. Thus the radiation beam orientation was variable only in plane of the CT scan and the central ray could not be positioned at oblique angles with respect to the transverse image planes. Isodose lines were calculated at 10% intervals and subsequently plotted at full scale by incremental plotter. Finally, the isodose distribution for each CT section was entered in PDP 11/50 computer by tracing each isodose line with a graf-pen. The computer used the image scaling factors and coordinate systems as previously described to combine the CT scan with the isodose distribution by assigning an isodose value to the locus of pixels which were most spatially coincident with each isodose line. RESULTS A representative transverse scan with the isodose distribution resulting from a 45” wedge pair is shown in Fig. 1. Saggital (Fig. 2) and coronal (Fig. 3) views were reconstructed by requesting the computer to
November-December 1978. Vol. 4, No. 11 and
NO. 12
Fig. 2. Computer reconstructed saggital section of image and isodose curves.
Fig. 3. Computer reconstructed coronal section of image and isodose curves.
retrieve from each transverse section, the appropriate pixels which make up the desired plane. The computer displayed the CT scan in monochrome grey-scale while the isodose lines could be displayed either in polychrome or monochrome grey-scale (using a color different from that of CT scan by assigning a different color or shade to each isodose value). DISCUSSION One of the first uses of CT information for radiation therapy treatment planning was reported by Chernak et al.,’ who utilized CT scans for planning in single transverse planes. Several other authors’” have published similar results with the addition of greater utilization of CT data for inhomogeneity corrections. Initial results of the clinical efficacy of CT use in radiotherapy have been reported by Munzenrider et ~1.~In a recent report of the Committee on Radiation Oncology Studies Sub-committee on CT scanning and
Fig. 1. Computer display of treatment plan on a cross section at the level of the central beam.
Radiation Therapy6 sional calculation
the importance and display
of three-dimentechniques was
Study of computerized tomography radiation therapy treatment ?? R. AB~DIRet al.
emphasized. While true three dimensional radiation beam models do exist”’ the emphasis of most current treatment planning systems is on single or multiplanar calculation. Hence one approach, the one examined here, is to utilize multi-planar image and isodose information to generate coronal and saggital information. There are several problems with this demonstration system. A large number of scans were used. As shown previously,4 this close spacing was necessary in order to maintain a resolution in the reconstructed coronal and saggital planes which is equivalent to that in the transverse planes. The large number of scans is time-consuming and gives rise to motion and sequential orientation problems. These difficulties can be minimized by reducing the required number of scans and the time per scan. A fast CT scanner is required for a clinically feasible CT-RT system. In order to reduce the number of scans in the axial direction it is necessary to sacrifice spatial resolution of the CT image display of reconstructed planes and develop three dimensional treatment planning systems which can interpolate in all planar views. As previously noted, external beam treatment planning programs of most present radiotherapy computers interpolate only in single planar sections although 3D calculations are possible using an irregular field approach. True three-dimensional beam models and isodose interpolation based on a 1 cm cubic grid point
1109
system in conjunction with fast CT scanners probably would reduce the total scan time to approximately 2 min for 10 cm in the axial direction. The present system is limited to coronal and saggital planes which are othogonal to the transverse scan planes. Furthermore, it is feasible to position the beams only in planes parallel or perpendicular to the plane of transverse scans. A system that is clinically feasible should be able to position beams arbitrarily, to calculate isodose distributions, and to reconstruct pictures in any plane. Again this will require flexible three dimensional treatment planning software. The present system depends far too much upon operator action to define coordinate systems, orient scans, relative to each other and input data. The total time for a complete study, even after the CT scans are acquired, is at least 4 hr. This is not of concern for a feasibility study of this type; however it is unacceptable for a clinical system. Hence, for ease and speed of operation, it would be most convenient to have the image processing and treatment planning capability implemented on a single computer. This would allow a unified, simpler, approach to the solution of problems such as scaling, orientation of sequential sections, beam positioning, isodose calculations and display of the results. It would require a computer of greater capacity and speed than the one that presently is used exclusively for radiation therapy planning.
REFERENCES 1. Chemak, E.S., et al.: The use of computer tomography
for radiation therapy treatment planning. Radiology 117: 613-614, 1975. 2. Fullerton, G.D., et al.: CT determination of parameters for inhomogeneity corrections in radiation therapy of the esophagus. Radiology 126: 167-171, 1978. 3. Geise, R.A., McCullough, E.C.: The use of CT scanners in megavoltage photo-beam therapy planning. Radiology 124: 133-141, 1977. 4. Glenn, W.V. et al.: Image generation and display technique for CT scan data. Invest. Rad. 10: 403-446, 1975.
5. Munzenrider, J.E. et al.: Use of body scanner in radiology treatment planning. Cancer 40: 170-179, 1977. 6. Steward, J.R., Hicks, J.A., Boone, M.L.M., Simpson, L.D.: Computed tomography in radiation therapy. Int. J. Radiat. Oncol. Biol. Phys. 4: 313-324, 1978. 7. van de Geijn, J.: A computer program for 3-D planning in external beam radiation therapy, EXTDOS. Camp. Prog. Biomed. 1: 47-57, 1970. 8. Weinkam, J., Sterling, T.: A versatile system for the three-dimensional radiation dose computation and display, RTP. Comp. Prog. Biomed. 2: 171192, 1972.
