SPOT REGION OF INTEREST IMAGING: A NOVEL FUNCTIONALITY AIMED AT X-RAY DOSE REDUCTION IN NEUROINTERVENTIONAL PROCEDURES

Abstract Aim of the study: The aim of this study was to describe a new functionality aimed at X-ray dose reduction, referred to as spot region of interest (Spot ROI) and to compare it with existing dose-saving functionalities, spot fluoroscopy (Spot F), and conventional collimation (CC). Material and methods: Dose area product, air kerma, and peak skin dose were measured for Spot ROI, Spot F, and CC in three different fields of view (FOVs) 20 × 20 cm, 15 × 15 cm, and 11 × 11 cm using an anthropomorphic head phantom RS-230T. The exposure sequence was 5 min of pulsed fluoroscopy (7.5 pulses per s) followed by 7× digital subtraction angiography (DSA) runs with 30 frames per DSA acquisition (3 fps × 10 s). The collimation in Spot F and CC was adjusted such that the size of the anatomical area exposed was as large as the Spot ROI area in each FOV. Results: The results for all FOVs were the following: for the fluoroscopy, all measured parameters for Spot ROI and Spot F were lower than corresponding values for CC. For DSA and DSA plus fluoroscopy, all measured parameters for Spot ROI were lower than corresponding parameters for Spot F and CC. Conclusion: Spot ROI is a promising dose-saving technology that can be applied in fluoroscopy and acquisition. The biggest benefit of Spot ROI is its ability to keep the entire FOV information always visible.


INTRODUCTION
Endovascular treatment of cerebral vascular diseases has developed tremendously during the last 30 years, which has led to a change in treatment strategy and an improvement of the outcomes of treatment of these diseases. In addition, thanks to numerous newly developed endovascular methods of treatment, it is possible to treat some cerebral vascular diseases which were untreatable only 15 or 20 years ago. However, long fluoroscopic times and frequently repeated angiographies, necessary for obtaining highquality visual information, inevitably carry the risk of the skin damage caused by ionizing radiation (1)(2)(3)(4)(5)(6) .
Many systems aimed at direct or indirect X-ray protection have been developed. Among them, beam collimators of various forms and constructions were one of the first devices for reducing X-ray dose and have been used almost since the beginning of the Xray era (7,8) .
Though very effective, conventional collimation (CC) has considerable disadvantages: 1. Only symmetric collimation is possible, which leads to an unnecessary exposure for larger anatomical areas than actually needed. 2. Anatomical and/or device-relevant reference information is lost.
3. Patient skin dose is increased when collimating inside the automatic brightness control region of interest (ABC ROI) (7) .
CC uses the standard ABC technique with a constant flat panel detector dose, fixed ABC sensing area in size and position, as well as a blade rejection for brightness calculation (7) . Besides the CC, several other functionalities aimed at optimizing data acquisition and image post-processing and thus reducing the radiation dose have been developed during recent years.
Söderman et al. described a new system for optimization of data acquisition and image creation, which is integrated in the Philips 'Allura' biplane angiographic machine (9) . Han et al. showed that the use of special protection can lead to considerable dose reduction (10) .
According to Khan et al., dose reduction can be achieved despite prolonged intervention time if modification of default settings on biplane angiography equipment is applied (11) . Borota and Patz have shown that improved fluoroscopy based on a new construction of ABC as well as flexibility of lateral isocenter of a biplane angiographic machine significantly contributes to the reduction of the dose (12,13) . Figure 1. (a-c) Spot ROI. The brighter square in the center indicates the ROI exposed with a normal dose. The darker surrounding area indicates the area with additional attenuation of 0.7 mm Cu; (d-f) Spot F. The white frame in the center indicates the ROI exposed. Surrounding area is shielded by collimation with LIH superimposed; (g-i) CC The development of spot fluoroscopy (Spot F) some years ago has greatly improved the situation as it permits acentric, asymmetric collimation (12) . However, as it is still based on collimation, the entire field of view (FOV) anatomy is not visible, and some anatomical and/or device-relevant information is still hidden (Figures 1 and 2). Spot region of interest (Spot ROI), a novel functionality, has been developed and recently become available for reduction of X-ray dose during neurointerventional interventional procedures. The functionality, as well as Spot F, is integrated into the commercially available new 'Alphenix' biplane angiographic machine designed and manufactured by Canon Medical Systems (Canon, formerly Toshiba Medical Systems, Tochigi, Japan) (Figures 1 and 2  (e and f) Spot F with acentric, asymmetric collimation position the ROI always over the vascular structure of interest independently of its location within the FOV selected. Therefore Spot ROI provides always full FOV information. As the Spot ROI is of a fixed size, the anatomical area exposed is constant, but its displayed size varies over the different FOVs following the same geometrical magnification rules (Figure 1a-c). 2. Novel ABC technique which in contrast to the conventional ABC technique uses an adaptive ABC ROI instead of a static ABC ROI (14) . As the Spot ROI filter can freely be moved, the ABC ROI needs to accurately track the position of the Spot ROI in real time in order to ensure a correct brightness detection and interpretation, which is an essential precondition for the dosesaving effect. Since the backscatter is considerably reduced by the additional 0.7 mm Cu around the Spot ROI, the detector input dose is reduced by a certain percentage to maintain a signal-to-noise ratio similar to that achievable without Spot ROI and higher backscatter.
The aim of this study was to compare the dose impact of the novel Spot ROI functionality with the existing dose-saving techniques CC and Spot F.

MATERIAL AND METHODS
An anthropomorphic head phantom RS-230T was used as a target ( Figure 3). The system was equipped with a 30 × 30 cm a Si flat panel detector (TFP1200C/A1.OEM, Varian Medical Systems, CA, USA) with a pixel size of 194 μm.
Table height was 105 cm; source to image distance was 100 cm with the C-arm in PA orientation. The exposure sequence was 5 min of pulsed fluoroscopy (7.5 pulses/s), followed by 7× digital subtraction angiography (DSA) runs with 30 frames per DSA acquisition (3 fps × 10 s). The collimation in Spot F and CC was adjusted such that the size of the anatomical area exposed was as large as the Spot ROI area in each FOV ( Figure 2). Dose area product (DAP) was measured using the built-in DAP meter Diamentor K1/K2 (PTW Freiburg, Germany) for all fields of view.
The air kerma (AK) value is automatically calculated by dividing the DAP by the area exposed at the interventional reference point (IRP) and displayed on the system monitor. It is a standard built-in function of any angiographic system. Peak skin dose (PSD) was used to identify and display the patient skin surface area with the highest cumulative dose delivered. It uses system-specific calibration data and The accuracy tolerance of PSD estimation has been proven to be ∼10% which is in fact more accurate than the usual DAP meter accuracy with a tolerance of 35% (15) . The technical characteristics of pulsed fluoroscopy used for the measurements were detector input dose of 0.041 μGy per frame for a reference FOV of 20 × 20 cm and 0.061 μGy per frame for reference FOVs of 15 × 15 cm and 11 × 11 cm. The beam filter was 0.3 mm Cu for all fluoroscopy acquisitions. For the DSA series, the detector input dose was 1.45 μGy per frame for a reference FOV of 20 × 20 cm and 1.75 μGy per frame for reference FOVs of 15 × 15 cm and 11 × 11 cm. The beam filter was 0.2 mm Cu for all DSA acquisitions. The X-ray parameters for all acquisitions can be seen in Table 1. DSA acquisitions of the entire FOV for Spot F and CC were performed for two reasons: 1. To create a sufficient patient skin dose map that also permitted a clear visual difference in the total dose applied between the three modalities. 2. To simulate the clinical reality in which DSA is always used after or between fluoroscopic runs to obtain the entire FOV information. Therefore, no collimation was applied in Spot F and CC, while Spot ROI was applied because it always keeps the entire FOV information visible.

RESULTS
Our results are summarized in Tables 1-3. The size  Due to the essential impact of the dose generated by runs (angiographies) on the total dose, parameters of total dose (AK, DAP, and PSD) for Spot ROI were lower than the corresponding parameters for Spot F and CC regardless of the size of the FOV.

DISCUSSION
While Spot ROI and Spot F show about the same PSD values, AK and DAP differ markedly between the three different modalities. The PSD refers to the patient surface area of the highest cumulative skin dose, which in this case is the area of the central beam. As Spot F and Spot ROI show practically the same X-ray parameters, the same PSD is the logical consequence. Spot ROI shows the highest DAP value in fluoroscopy mode for FOV 20 and FOV 15 because the exposed field size is significantly larger for SR (Table 1) even though the X-ray intensity is different between the Spot ROI and the surrounding area.
For FOV 11 Spot ROI shows quasi the same DAP than Spot F which could be explained by the lower X-ray parameter used for SR as a result of the ABC response, the smaller difference in effective field size between Spot ROI and Spot F as well as the accuracy tolerance of the DAP meter response to the radiation.
The built-in DAP meter (PTW Diamentor) outputs only the DAP value.
The AK value is calculated by dividing the DAP by the area exposed at the IRP.
In fluoroscopy the area exposed with Spot ROI is ∼2.5-8× times larger than the area exposed in Spot F and CC, whereas the DAP difference between Spot ROI and Spot F, CC is only about a factor of 1.5 at max. Hence the resulting AK which takes the area exposed into account is markedly lower for Spot ROI in comparison with Spot F and CC. The difference between Spot F and CC with the identical area exposed can be explained by the different ABC methods used. Spot F uses a similar adaptive ABC technique than Spot ROI which is reflected in the respective X-ray parameters chosen by the system.
These results prove also the importance of a true patient model-based PSD computation and the limitation of AK-based dose values.
It shows that AK-based values can be misleading regarding the actual PSD hitting the patient as AK represents only an average dose value of a homogeneous air field exposed at the IRP, which is 15 cm below system ISO center toward the focus. While the IRP is a fixed point in space, the patient skin entrance plane in contrast is not. It varies with table height, patient thickness, angulation, and isocentric or non-isocentric positioning. Hence the patient skin entrance plane can consequently have either a larger or a smaller distance to the focus than the IRP or even exactly the same distance which results in an over-or underestimation of the PSD when AK is used as reference. Moreover, AK does not take into account PSD relevant and influencing factors such as patient tissue absorption and backscatter characteristics, absorption/scatter effects of table and mattress, the dose distribution pattern, as well as the differences in systems geometry. In conclusion the AK value displayed on the monitor does consequently not give a reliable and correct indication of the actual PSD delivered to the patient. That is why the PSD estimation model used in the Alphenix system is not based on AK as input value but uses a dedicated FDA-approved algorithm for dose computation which is completely separated from the traditional AK calculation using the DAP meter value as input.
The precondition for DSA was to get the entire FOV information. Consequently, in Spot F and CC, no collimation was applied i.e. the effective field size exposed was identical between Spot ROI, Spot F, and CC for all FOVs. CC uses conventional ABC principles like those used in fluoroscopy. The values for Spot F and CC are the same because both use the same ABC technique, resulting in the same Xray parameters and dose. In contrast, Spot ROI can be applied in all acquisition modalities (fluoroscopy, DSA, digital angiography) i.e. the novel ABC technique described above is also active in DSA. This, in combination with an additional 0.7 mm Cu layer, results in considerably lower dose values for each comparator concerned (AK, DAP, PSD).
The total dose value as the sum of fluoroscopy and DSA is consistent with Spot ROI being the method with the lowest values for AK, DAP, and PSD. Our results showed that Spot F and Spot ROI are superior to CC for each measured parameter regardless of the size of the FOV.
The main advantage of the Spot F in comparison to Spot ROI is that a rectangular or square ROI of any size can be freely moved within the FOV at any time and as often as the operator wants. The FOV, outside the ROI, is completely shielded by collimators. That means, however, that the FOV outside the ROI is hidden and the image displayed on the screen during the fluoroscopy is only the LIH, not a real-time image. The square ROI of constant size generated by Spot ROI can also be moved within the selected FOV. In contrast to the ROI generated by Spot F, the FOV outside this square ROI is still visible, making it possible to track in real time the most important anatomic or device-related information (Figure 2a-c).

Methodological limitations
There are three limitations of our study. First, statistical analysis of results was not performed since the data we have generated are not appropriate for a statistical evaluation.
We have only one pair of data for each FOV and both acquisition modes (fluoroscopy and DSA). Though the percentage differences between the measured parameters (AK, DAP, and PSD) are obvious, the calculation of p-value of these differences would not give meaningful results. Second, the results of our measurements were not compared to results of other studies because studies performed under the same experimental conditions and the same or a similar system have either not been carried out or are not available in the scientific literature. Third, an important limitation of this study is its experimental character. A new study, in a clinical setting, with live cases, is necessary to obtain reliable and comprehensive assessment of the practical benefits of this system. This will be possible since the Spot ROI is integrated into the commercially available 'Alphenix' biplane angiographic system.
Finally, evaluation of the Spot ROI under clinical conditions with large number of patients, as already described in the literature with another dose-saving functionality (9) , would enable statistical analysis of results which is missing in this study.

CONCLUSION
Spot ROI is a promising dose-saving technology as it can be applied in fluoroscopy and DSA and digital angiography acquisition. The biggest benefit of Spot ROI is its ability to keep the entire FOV information always visible. Despite limitations, a combination of the use of Spot F and Spot ROI, depending on the clinical situation, is a potentially useful technique for obtaining an appropriate quality of visual information while reducing the radiation dose received by the patient.

CONFLICT OF INTEREST STATEMENT
The authors declared the following potential conflicts of interest with respect to the research, authorship, and/or publication of this article: The unit for neurointervention in our department is Canon's reference site. AP is a Canon employee, an engineer, and International Clinical Development Manager.