[A method for optimizing the estimation of the dose to the patient in traditional radiology]

The method recommended by Report no. 34 (1982) of the International Commission on Radiological Protection (ICRP) for patient dose computation in diagnostic radiology is based on tabulated dosimetric data obtained from Monte Carlo simulations on anthropomorphic phantoms described by simple mathematical functions. When computing the dose absorbed by an adult patient, this method suffers from two main limitations: first, the geometrical parameters--and in particular focus-to-film distance and film size--are fixed, which makes the dosimetric data of limited use when the examination geometry differs from the ICRP standard. In addition, when patient size and mass differ considerably from the corresponding quantities of the mathematically described phantom (the so-called reference man, with a height of 174 cm and a mass of 70.9 kg) the ICRP method may lead to great errors in dose estimate. The aim of the present paper is to indicate a method to overcome the above limitations.

The algorithm proposed in this work is based on the method suggested by Huda and Gkanatsios in order to compute the effective dose through a linear first of the energy imparted per unit dose-area product as a function of the half value thickness and by using fit coefficients depending on both phantom thickness and peak voltage. We devised a procedure to normalize the dose computed with this methods with respect to the equivalent effective dose obtained with the ICRP method. We therefore determined the dependence of the absorbed dose on focus-to-film distance, film size and patient anatomy.

We found that--for each value of patient mass--the dose dependence on film size can be approximated by a polynomial function, while the dose dependence on focus-to-film distance can be approximated by a power law. If the above parameters vary in a limited range close to the ICRP standard, a linear fit can be performed without introducing a considerable error. The linear fit coefficients, on the other hand, were found to depend on the average body surface, a parameter which takes into account both patient height and mass. Thus, determining the normalization factor for each projection and each view allows to estimate the absorbed dose under different geometrical conditions. The method has been verified by considering four of the most common X-ray procedures (chest AP, cervical spine LAT, lumbar spine AP and head LAT).

The average error on dose estimation is about 13%. In the very next future the method will be extended to all the projections and views of ICRP Report no. 34, and we plan to integrate the described algorithm in a computer program devoted to the automatic computation of patient dose.