PATIENT SIMULATION IN X‐RAY THERAPY *

Radiation therapy remains one of the two primary treatments for cancer, the other being surgery. Ionizing radiation in large amounts is always damaging to tissue. The problem faced by the therapist is to deliver a lethal dose to cancer tissue, while not damaging normal tissue past the point where it can repair itself; or (more generally) to devise a plan for radiation treatment which will maximize curative or palliative effect while minimizing undesirable systemic effects. A given treatment is most, easily evaluated by preparing a chart which shows, on a cross section of the patient,. the distribution of radiation dose in the form of "contours" of equal radiation dose, or isodose curves. Such charts are usually prepared by graphically combining known curves for single fields. This involves considerable manual labor by trained personnel and is seldom done except in large institutions, and even then for only a small proportion of the patients, with perhaps one or two treatments beng plotted for each patient. Automatic computation of treatment plans makes possible the calculation of several alternative treatments for each of" many patients, and greatly aids the therapist in determining an optimal treatment plan. These advantages were recognized at Memorial Hospital about 10 years ago, and the first paper describing the Memorial Hospital automatic isodose calculation system was published in 1955.' The equipment then used was limited to a keypunch, sorter, and accounting machine. In 1959, a new mathematical model for describing the treatment was developed which permits simulating the treatment of the patient on either a digital or an analog computer. This model was published in 1962,2 but since it differs substantially from other models by Sterling,":' Hallden, Perez-Tamayo, Wood, and Cunningham," the rationale of its development will be described here. In this development, use was made of two concepts: first, a vector u is equivalent to a digitized function of a single variable f(x), and a matrix A is equivalent to a digitized function of two variables g(x,y); second, if a matrix A results from digitizing a function of two variables f(x,y) at equal increments in x and equal increments in y, and if A is of rank n, then: