In 1995 we celebrated the centennial of Roentgen's initial description of the x-ray. Spurred on by Roentgen's remarkable discovery, investigators immediately began to determine if any naturally occurring substances also emitted penetrating radiation. This led to the discovery of radioactivity by Becquerel in 1896 and to the isolation of radium by Marie and Pierre Curie in 1898. The fact that radiation, either from x-ray tubes or from radioactive isotopes, had a biologic effect was discovered almost immediately as early investigators began to show radiation-induced skin changes, most frequently on their hands. Many of the early pioneers of radiology later died of metastatic cancer caused by unshielded exposure to their research.
The therapeutic use of radiation followed shortly after its discovery. It is a matter of some debate as to who was ``first'' to use radiation in an attempt to cure cancer. Although conflicting claims abound, it is undeniable that radiation was being used to treat superficial tumors, such as skin cancers, prior to the dawn of the twentieth century. The technology of x-ray tube production improved rapidly, and by the 1920s equipment with reliable and reproducible x-ray output was widely available. The first standardized unit of radiation measurement, the roentgen, was described in the late 1920s, and the clinical science of radiation therapy was poised to make enormous advances.
Throughout the first half of this century, radiation beams were predominantly of low energy, in the range of 150 to 300 kV (a typical chest x-ray film is taken with x-ray energies of 80---100 kV). Although these orthovoltage energies could penetrate fairly deeply into tissue, they deposited their maximum energy directly on the surface of the skin. This was ideal for the treatment of superficially located lesions but presented a serious drawback to the treatment of deep-seated masses. For substantial doses of radiation to be applied to deeply located tumors, it was necessary for the radiation to be cross-fired from a number of different angles. The beams were aligned so that they would intersect inside the tumor-bearing region. This allowed the deposited beam energy to summate inside the tumor, while the dose to the more superficial structures was spread out among the several cross-firing beams.5 However, in order to have all the beams intersect properly within the target area, one needed to be able to plan the beam trajectories and to calculate the dose that would ultimately be achieved inside the tumor from a particular arrangement of beams. Thus, the science of radiation therapy treatment planning was born.
In the 1950s, radiation therapy entered the megavoltage era. With the advent of the cobalt machine, radiation oncologists had, for the first time, a reliable beam of over one million electron volts (MeV). Such a beam deposited its maximum dose 5 mm below the skin; thus, the skin itself no longer became the dose-limiting structure for each and every treatment plan. These higher-energy cobalt machines also added a new feature to the field of radiation oncology: isocentric treatment. The head of the machine was attached to a gantry that could rotate 360Æ around the patient. The point of rotation was fixed in space, usually 80 cm from the cobalt source, and no matter what angle the gantry was turned to, the center of the radiation field would always pass through this point of rotation (because it was always at the center of the field, this point was referred to as the isocenter). By placing the isocenter at the geometric center of the target, the beam would always intersect the target regardless of the gantry angle, thus facilitating the use of multiple cross-firing radiation fields.
In the 1960s, linear accelerators became commercially available, and these machines produced even higher radiation energies, from initial levels of 4 and 6 MeV to current levels of 25 MeV and more. Their isocenters were now at 100 cm, allowing greater freedom to rotate the machine to different angles around the patient. Doses of radiation were not limited by the tolerance of superficial structures, but rather by the tolerance of deep-seated normal tissues such as lung, kidney, liver, and spinal cord. The need for careful radiation therapy treatment planning became even greater.
The first radiotherapy treatment plans were performed by taking a manual contour of the outer surface of the body. This was typically done by molding a piece of solder wire using the outer shape of the body as a template or allowing a thin plaster strip to dry on the skin's surface. The solder wire or plaster strip was carefully removed and placed on a large sheet of paper, and the outer contour of the body was traced onto the paper. By transposing the location of normal structures and target regions from x-ray films taken at different angulations (typically at right angles to one another), an approximation of the location of important structures could be placed within the external body contour. This process actually created the first cross-sectional ``images'' of the body. To perform treatment planning and dose calculations, the dose deposition patterns of dozens of individual beams were maintained in an atlas. When a beam of specific size and direction was needed, a plot showing regions of equal dose as a series of lines (an ``isodose'' plot) for that beam was removed from the atlas and laid underneath the paper containing the cross-sectional drawing of the patient, and the isodose lines were traced by hand onto the drawing. Additional beams from additional angles were then traced, and where the individual beam isodose lines crossed, the dose was added together. In this way, a treatment plan was derived. This process was obviously slow and time-consuming work, and it was unlikely for planners to iterate through numerous beam arrangements looking for an ``optimal'' solution to a particular treatment planning problem.
Computerized treatment planning began in the late 1960s and dramatically enhanced the speed of radiotherapy treatment planning. The beam profiles were now stored in the computer and could be placed onto the patient contour quite rapidly. The early computers could sum the dose from numerous beams within a minute or two. A series of refinements to the plan could be made relatively quickly and easily until a final plan was created. However, treatment planning was still done on a single cross- sectional slice that was hand-drawn and of questionable accuracy. This one slice, usually through the center of the target volume, was treated as though it were representative of the entire therapeutic problem.
With the arrival of the whole-body CT scanner, radiotherapy treatment planning changed dramatically. Almost overnight, the hand-drawn cross-sectional images of the patient that had formed the basis for radiotherapy treatment planning for many decades were replaced by cross-sectional CT images. These images were geometrically accurate and displayed the location of the patient surface, normal tissues, and target structures with unprecedented clarity. In addition, multiple cross-sectional images at many levels within the patient were readily available. It did not take radiation oncologists long to begin to use these CT slices as the basis for radiotherapy treatment planning. In the late 1970s and early 1980s, a series of reports confirmed the utility of CT-based treatment planning. Compared with conventional planning, CT information caused significant changes in planning technique in one third to two thirds of patients studied, depending on the anatomic location of the tumor. In addition to showing us detailed anatomic information, CT was the first imaging modality to produce information concerning the density of tissues. Because the attenuation of radiation is dependent upon the electron density of the tissue being traversed by the beam, correcting for these tissue inhomogeneities allowed far more accurate dose calculations to be performed. In less than a decade, CT-based treatment planning replaced manual treatment planning in a substantial number of anatomic sites. However, treatment planning was still performed on a single cross-sectional slice through the center of the tumor or, at times, with the addition of one or two slices above and below the central plane.
Radiologists and radiation oncologists recognized that a CT scan was a consecutive series of two- dimensional cross-sectional slices that made up a full three-dimensional (3-D) description of the patient. By ``stacking'' the CT scans one on top of the other, the inherent three-dimensional nature of this imaging modality could be displayed. By the mid-1980s, diagnostic radiology systems producing fully 3-D anatomic renderings were becoming commercially available. The use of 3-D reconstructed images for radiotherapy treatment planning was also being developed at the same time. The process, however, was far more complex in radiation oncology than it was in diagnostic radiology, for the following reasons:
1. Images used for treatment planning had to be geometrically precise, because distortions involving the shape, size, or location of structures could dramatically affect dose calculations and potentially lead to complications.
2. One had to be able to display radiation beams on 3-D anatomic structures in 3-D space. This meant that the beams had to be displayed with their full 3-D divergence, and one needed to be able to ``slice'' the 3-D display along any arbitrary plane to see the intersection of the radiation beam with objects on that plane.
3. One had to be able to calculate and display a radiation dose in three dimensions, taking into account the 3-D influence of tissue inhomogeneities and scatter. This meant that one had to create software that could recognize when anatomic objects were within or outside a beam path, create computer models of dose distribution in three dimensions, and calculate and display 3-D dose quickly and accurately.
Although these problems were formidable, they were able to be overcome, and 3-D radiotherapy treatment planning systems were created in a number of institutions in the United States and abroad.
PRINCIPLES & PRACTICE OF ONCOLOGY / May 1994, Number
5 /Allen S. Lichter, M.D.
Randall K. Ten Haken, Ph.D.