**1.3. Protection against radiation exposure**

Short after the discovery of ionizing radiation by Röntgen in 1895, its detrimental effects became apparent, and people tried to protect themselves [24]. Nowadays, the International Committee on Radiological Protection (ICRP) and the US National Council on Radiation Protection and Measurements (NCRP) aim to protect people by advising means for achieving this, e.g., regulatory and guidance limits [18, 25].

The major question that keeps radiation protection bodies busy and that became the foundation of radiation protection guidelines worldwide is "How much is harmful?" This question is particularly relevant for low-dose exposures for which health impact is not yet fully elucidated. Although a large number of epidemiological and radiobiological studies have been performed to date in order to investigate the effects of low doses of ionizing radiation [26–47], accurate risk assessment is not yet available [18]. Current guidelines for protection against low-dose radiation are based on cancer risk estimates from epidemiological studies. As discussed further, cohorts include atomic bomb survivors, occupationally exposed people, patients (diagnostics or therapeutics), and environmentally exposed people [48]. In general, an excess cancer


\* Doses are whole-body doses, except those of medical exposure, which are delivered to a specific organ. CT, computed tomography; Sv sievert [7, 19–23]

**Table 1.** Representative effective doses associated with different sources of ionizing radiation.

risk can be statistically evidenced for doses above 100 mGy. Nevertheless, doses below 100 mGy are inconclusive due to two practical limits of epidemiological studies: low statistical power that generates random errors and demography that gives rise to systematic errors. Due to a high natural incidence of cancer, a lifetime follow-up of larger cohorts would be needed to quantify excess cancer risks due to a low dose of ionizing radiation. This is practically infeasible. Furthermore, confounding factors, such as lifestyle risk factors for CVD, can hamper accuracy to confidently detect a small increase in cancer mortality (discussed in Section 2.4). Any inadequacy in matching between control and study groups may give rise to a bias that cannot be merely reduced by expanding the size of the groups [49]. As a consequence, risk assessment in the low-dose region (<100 mGy) is based on extrapolations made from high-dose risk estimates [50]. For cancer, it is widely accepted that the tumorigenic risk increases with radiation dose without the presence of a threshold (the stochastic linear nonthreshold [LNT] model). This assumption implies that no dose is absolutely safe, resulting in implementation of the "as low as reasonably achievable" principle [51, 52].

In contrast to cancer, non-cancer diseases have for long not been considered as health risks following exposure to low doses of ionizing radiation. Consequently, they were believed to have a threshold dose below which no significant adverse risks are induced (deterministic linear threshold model) [18, 53]. This idea has been challenged by epidemiological findings showing an excess risk of non-cancer diseases following exposure to doses lower than previously thought [34, 54]. Epidemiological evidence suggests an excess risk of CVD mortality above 0.5 Gy [34, 54]. For doses below 0.5 Gy, the dose-risk relationship is still unclear. However, if the relationship proves to be without a threshold, this may have a considerable impact on the current radiation protection system, since the overall excess mortality risk following low-dose exposure could double [55].
