Task-based measures of image quality and their relation to radiation dose and patient risk

Abstract

The theory of task-based assessment of image quality is reviewed in the context of imaging with ionizing radiation, and objective figures of merit (FOMs) for image quality are summarized. The variation of the FOMs with the task, the observer and especially with the mean number of photons recorded in the image is discussed. Then various standard methods for specifying radiation dose are reviewed and related to the mean number of photons in the image and hence to image quality. Current knowledge of the relation between local radiation dose and the risk of various adverse effects is summarized, and some graphical depictions of the tradeoffs between image quality and risk are introduced. Then various dose-reduction strategies are discussed in terms of their effect on task-based measures of image quality.

Figure 1

ROC curves. Left: Single curve showing three different operating points A, B and C, corresponding to three decision thresholds t_c. Right: Three different ROC curves, corresponding to different values of the detectability index d_A (hence different AUCs).

Figure 2

Left: Schematic Localization ROC (LROC) curve. Right: Schematic Estimation ROC (EROC) curve.

Figure 3

Schematic Quality-Dose Characteristic (QDC) curves for a binary classification task. Left: Quality specified by SNR². Right: Quality specified by area under the ROC curve (AUC). For both panels D can be any suitable dose measure (see text).

Figure 4

Schematic curves of risk vs. dose for three response models. The term ‘risk’ is used broadly here and in subsequent graphics to include any of the absolute or relative risks defined in Sec. 5 or the rates of cancer incidence given by (5.4) or (5.5). The solid curve is the conventional linear response model with no threshold (LNT); the dashed curve is the linear-quadratic (LQ) model of (5.6), and the dotted line is a linear-quadratic model modified with an exponential factor (LQE) as in (5.7).

Figure 5

Schematic Quality-Risk Characteristic (QRC) curves. Left: SNR² on a binary detection task for each of the three risk models shown in Fig. 4 and for NPWMF and Hotelling observers. Right: AUG for the same observers and risk models.

Figure 6

Plots of the Weibull probability of a tissue-reaction effect. Left: threshold set at 1% incidence as recommended in [148] and various shape factors. Right: Fixed shape factor and various thresholds. Note that the graph on the right covers a small area in the lower left corner of the graph at the left.