From a clinical point of view, our example is also relevant as, brain tissue is at certain risk for late tissue reactions and (deterministic) effects [
40]. Although not common, in complex interventional procedures brain radiation doses above the ICRP absorbed dose threshold of
\( 500\,\hbox {m}\hbox {Gy}\) have been reported [
41,
42]. In addition, focusing on a large and mostly homogeneous organ gives an interesting comparison between measuring discrete dose values and simulating continuous dose distributions. We used the same X-ray system (Artis zeego, Siemens Healthcare GmbH, Forchheim, Germany) for all measurements. In total, we covered three standard neuroradiology projection angles,
\( 0 ^{\circ }\),
\( 45 ^{\circ }\), and
\( 90 ^{\circ }\), relating to posteroanterior, oblique, and lateral view directions [
43]. Furthermore, two peak tube voltages (
\( 70\,\hbox {k}\hbox {V}\)p,
\( 1.31\,\hbox {mm}\) Al air kerma half-value layer, and
\( 90\,\hbox {k}\hbox {V}\)p,
\( 1.68 \hbox {mm}\) Al air kerma half-value layer) were used in our experiments. The C-arm’s isocenter was aligned with the center of the phantom’s head, the source-to-isocenter distance was
\( 80 \hbox {cm}\), and the source-to-image distance was
\( 120\,\hbox {cm}\), respectively. For each imaging setting, we irradiated the phantom with
\( 100\,\hbox {m}\hbox {A}\) tube current for
\( 20\,\hbox {s}\) with 30 frames per second to ensure sufficient exposure of all MOSFET probes; neither pre-filtration nor collimation was applied. To account for the MOSFET uncertainty and to ensure stable average dose values for each measurement point, we repeated each acquisition five times. After each irradiation, we waited
\( 5\,\hbox {min}\) to ensure total discharge of the MOSFET probes. Therefore, the measurement protocol took
\( 26\,\hbox {min}\) and
\( 40\,\hbox {s}\) (including
\(5\times 20\hbox { s}\) acquisition time) for one imaging setting, leaving sufficient time to run and finish our simulation in parallel (online). The average air kerma was
\( 49.28\pm 0.07 \hbox {m}\hbox {Gy}\) for the
\( 70\,\hbox {k}\hbox {V}\)p spectrum and
\( 84.66\pm 0.10\,\hbox {m}\hbox {Gy}\) for the
\( 90\,\hbox {k}\hbox {V}\)p spectrum, respectively. The MOSFET probes were placed inside the ATOM phantom as shown in Fig.
4. To affix the MOSFET probes, we encased them in soft tissue-equivalent holders of the same size and shape of the drillings in the phantom. Since this experiment focused on the overall agreement of the computational and the experimental approach, we carried out the associated simulations offline using
\(25 \times 10^{8}\) primary photons and the high-resolution digitized phantom.