0 Editorial
COMPUTED
TOMOGRAPHY TREATMENT
IN RADIATION PLANNING
THERAPY
WILLIAM R. HENDEE, Ph.D. Department of Radiology, University of Colorado Medical Center, 4200 East Ninth Avenue, Denver, CO 80262, U.S.A. Computed tomography,
Radiation therapy treatment
planning.
In radiology, as in most fields of medicine, one can remember the evolution of elegant solutions in search of a problem. A few years ago sophisticated infrared mapping techniques that were developed primarily for military uses were thought to be immediately Unapplicable to detection of breast cancer. fortunately, clinical experience has proven otherwise, and most investigators who were interested in thermography now are back at the drawing board. An example closer to radiation oncology involves a number of institutions which invested in automatic isodensity plotters to obtain composite isodose distributions for megavoltage photon beam treatments, only to find such plotters relatively useless except for simple applications such as single field electron dose distributions. One would hope that experiences of this type would teach us to evaluate technological breakthroughs carefully in terms of their applications to radiology before climbing aboard a bandwagon. In radiation oncology, our next challenge to this improved perception is evolving in the area of the applications of computed tomography (CT) to radiation therapy treatment planning. Specifically, the question to be addressed is the value of automatic entry of electron density cross sections into megavoltage photon beam software packages of treatment planning computers for correction of dose distribution for the presence of body inhomogeneities. This question has been raised in discussions with many manufacturers of whole body CT units, and is implicit in a number of recent articles in the literature, including the article by Kijewski and Bjarngard in this issue of the Journal (see pages 429-435). A number of questions should be answered before the automatic entry of matrices of electron density data or CT numbers becomes another soughtafter and expensive option for whole body CT units used
wholly
or
in
part
for
megavoltage
treatment planning. Among these questions are: 1. Can improvements be made in present computational techniques to correct photon dose distributions for perturbations introduced by body inhomogeneities; can these improvements account for subtle influences such as the effects’ of inhomogeneities on scattered radiation without requiring computer memories and computational times much larger than those presently available with dedicated treatment planning systems? As Geise and McCullough’ have shown, present photon beam inhomogeneity correction methods can tolerate uncertainties in electron densities of 10% and in distances of 5 mm with errors no greater than 2% in corrected dose distributions. Can not this level of accuracy be met simply by using a digitizing device to trace in the inhomogeneities from a computed tomographic display sevice, and by assigning some average value or set of values for the electron density within the inhomogeneity without requiring an element by element knowledge of the electron density distribution? 2. With improved dose perturbation correction techniques, will not the accuracy of dose distributions be determined primarily by how well the patient geometry used to collect CT data simulates the patient geometry employed for radiation therapy? Also, will not other possible problems have to be addressed before automatic entry of CT data into treatment planning programs can be assumed to provide more accurate dose distributions? Among these potential problems are: inadequate beam hardening correction techniques; limitations in the frequency response of reconstruction algorithms; alignment difficulties and patient motion artifacts; and statistical fluctuations in CT numbers. One would hope that present uncertainties in the magnitude and clinical significance of improvements
therapy 539
540
Radiation Oncology ?? Biology 0 Physics
in radiation therapy treatment planning accompanying the direct use of CT data will be resolved both theoretically and experimentally before commercial organizations rush headlong into developing options
May-June 1978, Vol. 4, No. 5 and No. 6
for automatic entry of electron density cross sections into treatment planning packages for megavoltage photon radiation therapy.
REFERENCES 1. Geise, R.A. McCullough, E.C.: The use of CT scanners planning. in megavoltage photon-beam therapy
Radiology
124; 133-141, 1977.
COMPUTED
TOMOGRAPHY TREATMENT
IN RADIATION PLANNING
THERAPY
WILLIAM R. HENDEE, Ph.D. Department of Radiology, University of Colorado Medical Center, 4200 East Ninth Avenue, Denver, CO 80262, U.S.A. Computed tomography,
Radiation therapy treatment
planning.
In radiology, as in most fields of medicine, one can remember the evolution of elegant solutions in search of a problem. A few years ago sophisticated infrared mapping techniques that were developed primarily for military uses were thought to be immediately Unapplicable to detection of breast cancer. fortunately, clinical experience has proven otherwise, and most investigators who were interested in thermography now are back at the drawing board. An example closer to radiation oncology involves a number of institutions which invested in automatic isodensity plotters to obtain composite isodose distributions for megavoltage photon beam treatments, only to find such plotters relatively useless except for simple applications such as single field electron dose distributions. One would hope that experiences of this type would teach us to evaluate technological breakthroughs carefully in terms of their applications to radiology before climbing aboard a bandwagon. In radiation oncology, our next challenge to this improved perception is evolving in the area of the applications of computed tomography (CT) to radiation therapy treatment planning. Specifically, the question to be addressed is the value of automatic entry of electron density cross sections into megavoltage photon beam software packages of treatment planning computers for correction of dose distribution for the presence of body inhomogeneities. This question has been raised in discussions with many manufacturers of whole body CT units, and is implicit in a number of recent articles in the literature, including the article by Kijewski and Bjarngard in this issue of the Journal (see pages 429-435). A number of questions should be answered before the automatic entry of matrices of electron density data or CT numbers becomes another soughtafter and expensive option for whole body CT units used
wholly
or
in
part
for
megavoltage
treatment planning. Among these questions are: 1. Can improvements be made in present computational techniques to correct photon dose distributions for perturbations introduced by body inhomogeneities; can these improvements account for subtle influences such as the effects’ of inhomogeneities on scattered radiation without requiring computer memories and computational times much larger than those presently available with dedicated treatment planning systems? As Geise and McCullough’ have shown, present photon beam inhomogeneity correction methods can tolerate uncertainties in electron densities of 10% and in distances of 5 mm with errors no greater than 2% in corrected dose distributions. Can not this level of accuracy be met simply by using a digitizing device to trace in the inhomogeneities from a computed tomographic display sevice, and by assigning some average value or set of values for the electron density within the inhomogeneity without requiring an element by element knowledge of the electron density distribution? 2. With improved dose perturbation correction techniques, will not the accuracy of dose distributions be determined primarily by how well the patient geometry used to collect CT data simulates the patient geometry employed for radiation therapy? Also, will not other possible problems have to be addressed before automatic entry of CT data into treatment planning programs can be assumed to provide more accurate dose distributions? Among these potential problems are: inadequate beam hardening correction techniques; limitations in the frequency response of reconstruction algorithms; alignment difficulties and patient motion artifacts; and statistical fluctuations in CT numbers. One would hope that present uncertainties in the magnitude and clinical significance of improvements
therapy 539
540
Radiation Oncology ?? Biology 0 Physics
in radiation therapy treatment planning accompanying the direct use of CT data will be resolved both theoretically and experimentally before commercial organizations rush headlong into developing options
May-June 1978, Vol. 4, No. 5 and No. 6
for automatic entry of electron density cross sections into treatment planning packages for megavoltage photon radiation therapy.
REFERENCES 1. Geise, R.A. McCullough, E.C.: The use of CT scanners planning. in megavoltage photon-beam therapy
Radiology
124; 133-141, 1977.
