Lukmanda Evan Lubis Doctoral Thesis Summary

0,6 PMMA phantom 0,8 Plastic water phantom 0,5 PMMA-equivalent test material 0,7 Water-equivalent test material 0,4 0,6 0,3 (a) 0,5 (b) 0,2 0,4  (cm-1) 0,1  (cm-1) 0,3 0,0 0,2 30 35 40 45 50 55 0,1 30 35 40 45 50 55 25 Mean photon energy (keV) 0,0 Mean photon energy (keV) 25 50 (c) 50 (d) PMMA Water 40 40 30 30  (%)  (%) 20 20 10 10 0 30 35 40 45 50 55 0 30 35 40 45 50 55 25 Mean photon energy (keV) 25 Mean photon energy (keV) Figure 4.1. Resulting linear attenuation coefficient () measurements for (a) PMMA-equivalent material versus PMMA and (b) water-equivalent material versus plastic water as in the CIRS062M phantom. Deviations in linear attenuation coefficients as a function of energy for (c) PMMA- equivalent material versus PMMA, and (d) water-equivalent material versus plastic water as in the CIRS062M phantom. 4.2 Study on Dose and Image Quality Effects of Phantom Geometry 4.2.1 Dose index coefficients Measurements and calculations for three elliptical phantoms by using all tube voltages values were 0.39 ± 0.01 for C1, 0.16 ± 0.03 for C2, 0.31 ± 0.08 for C3, and 0.15 ± 0.03 for C4. Error bars of these coefficients are statistical errors from multiple measurements propagated over calculation results from all tube voltages. 42

From the results of dose coefficients there is no indication that the coefficients change with the variations of the phantom size and elliptical aspect ratio. Moreover, it should be noted that C2 and C4 applied on the measurement positions of AP side of phantom (i.e. 12 o’clock and 6 o’clock) have indistinguishable values with both being in the error bar of each other. This finding confirms that no separated coefficients are required for these positions and that patient table has little to no effect on dose measurement at 6’o clock position. It is of particular interest to observe if this trend applies for every beam energy (i.e., tube voltage). A particular interest should be drawn to C3 (lateral positions) with large error bars. From all results, only C3 indicates an irregular trend over different tube voltages. These trends can be explained by the scatter probability difference over different phantom sizes and aspect ratios. It can be further deducted that for lateral positions the effect of tube voltage is also influenced by the object size and aspect ratio. Care must be taken when applying these coefficients by considering the phantom size and aspect ratio. 4.2.2 Image quality trend The noise homogeneity values, H, from SNR measurements on all phantoms are presented in Figure 4.2, while results of MTF measurement are displayed in Figure 4.6. The SNR results were separated into central SNR (Figure 4.4), lateral and AP SNR to enable further discussion on the influence of phantom size and shape to noise level. From Figure 4.2, the homogeneity of the cross-sectional noise level has a slight trend of increase with the increase of the object size, i.e., phantom effective diameter. As commonly understood, larger objects create more absorption and therefore scatter, and the central axis, constantly being the most distant position from the X-ray tube, will receive less photons than the peripheral area. Thus, although no mathematical trend is observed, the result confirms that larger phantoms tend to have less homogenous noise. 43

0,35 0,30 80 kVp 100 kVp Noise homogeniety, H (CoV) 120 kVp 0,25 140 kVp 0,20 0,15 0,10 0,05 0,00 15 20 25 30 35 Phantom efective diameter, Deff (cm) Figure 4.2. Noise homogeneity values for all phantoms size and shape. 1,0 80 kVp 0,9 100 kVp 120 kVp 140 kVp 0,8 MTF 10% 0,7 0,6 0,5 0,4 14 16 18 20 22 24 26 28 30 Phantom effective diameter, Deff (cm) Figure 4.3. Results on MTF 10% measurement from all phantoms size and shape. 44

SNR1 16 14 80 kVp 100 kVp 12 120 kVp 140 kVp 10 8 6 4 2 0 15 20 25 30 35 Phantom effective diameter, Deff (cm) Figure 4.4. Central SNR trend as a function of phantom effective diameters. It is also important to highlight that there is no observable trend between phantom size (as well as aspect ratio) and spatial resolution. As shown in Figure 4.3, all MTF values are within the statistical errors of each other. This further confirms that the object size does not afflict the system’s capability of displaying image sharpness. Furthermore, from Figures 4.7, it can be concluded that the change of the object shape and size does not alter the trend of SNR as they all decrease exponentially with the increase of phantom diameter or effective diameter. In addition, the decrease is observed to be steeper with higher tube voltage. It can also be taken as confirmation that while the image quality metric was impacted by object size and shape, its trend (decrease over increasing size) tends to be independent of the object-related aspects and depends almost completely on the exposure parameters. However, the magnitude of image quality, especially noise, is afflicted by the phantom scatter volume. From the two aspects observed in this part of study, conclusions can be drawn as follows. In terms of dose, the coefficients used in elliptical phantoms are present and independent of the phantom size as well as tube voltage. When these coefficients are to be used for elliptical phantoms, consideration needs to 45

be taken for the use of C3 as coefficients for measurement in lateral positions. On the image quality side, there is no observable change of trend for MTF, SNR, and noise homogeneity with the change of the phantom size and shape. The change, however, is observable in noise homogeneity magnitude. Although the trend of these image quality metrics shift is associated only with the change of exposure parameters and not with cylindrical object aspect ratio, the scatter volume does influence noise, as theoretically stated. It can finally be concluded that producing or using elliptical phantoms for dose measurement and image quality assessment purposes requires special considerations on the size and aspect ratio of the phantom. 4.2.3 Considerations for the in-house phantom With these results and conclusions at hand and since the subsequent studies will be focused on cranial application, it was decided to use the 16 cm diameter, round-shaped phantom in the subsequent parts of the dissertation. Other phantoms will be dedicated to subsequent body 3DRA studies, which are not currently the focus of the dissertation. Moreover, it should be noted that 3DRA differs in nature with CT. With the use of large beam collimation and non-full single gantry rotation, the dose distribution is different. The CTDI formalism can therefore not be implemented in 3DRA. In an unpublished preliminary study by Hidayat et al (2021) using Monte Carlo simulation, the CTDI can be modified into use in 3DRA only after the coefficients are modified. It is only after then that the measurement results in the four peripheral positions will represent the averaged cross-sectional dose. This subject is currently reserved for separate future study. 4.3 Phantom Design, Construction, and Evaluation Results After the phantom material and geometry were decided from the first and second study parts, respectively, the phantom was manufactured and tested according to the methods described in Chapter 3. The results are described in this section. 46

Dose reading (mGy)4.3.1 Dose measurement reading Figure 4.5 shows the dose measurement result from the use of the two phantoms (in-house and standard CTDI phantom). In general, measurement in all positions did not not demonstrate major discrepancy. For air kerma reading, the in-house phantom has greatest absolute discrepancy against the reference standard phantom of 8.89% (3 o’clock position), with least discrepancy of 1.12% (6 o’clock position). Individual discrepancies are shown in Table 4.5. The national compliance test regulation issued by BAPETEN dictates that the dose measurement results should not differ by 20% in order for a CT modality to be certified. This difference refers to the discrepancy between measured and console displayed CTDIvol values. The fact that the constructed phantom differs only by less than 10% from a standard phantom indicates that the constructed phantom can be used for dose measurement purposes without indication that the results may alter the decision on the CT scanner’s state of performance. 8 Standard Head CTDI In-house Phantom 6 4 2 0 Center 12 o'clock 3 o'clock 6 o clock 9 o'clock Measurement position Figure 4.5. Dose reading from in-house phantom compared with standard CTDI phantom. 47

Table 4.5. Dose measurement result. Measurement Air kerma reading (mGy) Mean position discrepancy Standard CTDI In-house phantom (%) Center phantom 12 o’clock 4.51 ± 0.02 4.33 ± 0.07 4.01 3 o’clock 5.28 ± 0.12 5.21 ± 0.12 1.38 6 o’clock 4.67 ± 0.05 5.08 ± 0.34 8.89 9 o’clock 4.21 ± 0.26 4.16 ± 0.07 1.12 4.93 ± 0.17 4.78 ± 0.11 3.13 4.3.2 Image quality metric evaluation result The typical resulting image of the two phantoms are shown in Figure 4.11. The initial aim of the phantom construction was not to copy a commercially phantom. The tissue-mimicking objects on the electron density module were not the same as Catphan® phantom. Therefore, the evaluation was performed to check the linearity of the CT number—a parameter selected to represent electron density. Linearity is assessed by means of the r-squared values of the CT number. A similar approach was performed for contrast linearity by means of the SDNR. The in-house phantom was 4.66%, 5.56%, and 7.51% less linear than Catphan at 80 kVp, 100 kVp, and 120 kVp, respectively. This is mainly caused by the presence of two water- and adipose-mimicking objects originating from different raw materials (i.e., flour-based, FB, and carbon-based, CB). These two materials have demonstrated a difference in CT number and therefore are included in this study for linearity check. While the constructed phantom does not demonstrate a major contrast linearity difference against Catphan® 604, a major difference is observed in the contrast magnitude. With the Rose contrast model suggesting that human eye visibility threshold exists at SDNR = 5 (Burgess, 1999; Bushberg et al., 2011; Rose, 1973), the contrast level presented by the constructed phantom highly exceeds the threshold by a factor of ten compared to Catphan® 604. Therefore, threshold visibility research should take this into account. The CT number result of the insert objects compared with reference range (International Atomic Energy Agency, 2014) can be observed in Figure 4.7. 48

Figure 4.6. Typical resulting image of (a,b) electron density linearity and (c,d) contrast linearity checks obtained from (a,c) Catphan® 604 and (b,d) in house phantom at 100 kVp, 200 mAs. Images (a) and (b) are displayed with WL 10 and WW 1070, while (c) and (d) are displayed with WL 200 and WW 2500. It implies that the phantom module, produced with minimum volume of iodine contrast agent, provides higher detectability for assessment of contrast threshold and visibility. The result also suggested that the use of iodine contrast agent as solvent in resin provides superior detectability, although only in small volume. Furthermore, the correlation of phantom-based object contrast with clinical image contrast visibility requires further investigation. From Figure 4.8, it can be deduced that the thin wire on the constructed phantom can be used to measure 49

MTF with the resulting difference to Catphan® 604 being 2.00%, 4.45%, and 5.07% higher than Catphan® 604. The general result generally suggests that the designed and constructed phantom can be used for head CTDI measurements, CT number linearity assessment, as well as MTF measurements. Furthermore, the contrast linearity module, in addition to conformity with Catphan® 604, also suggests that the use of iodine contrast agent as solvent in resin provides superior phantom object detectability to corresponding Catphan® 604 module. This will be verified in Section 4.4.1. 100 50 CT number (HU) 0 -50 -100 80 kVp 100 kVp -150 120 kVp Reference -200 Liver Adipose (FABd)ipose (CB)Water (FBW) aterM(CusBc)le/brKaiidnn(egyreByrmaiantt(ewrh) ite matter) Figure 4.7. CT number result of insert materials compared with reference values. 50

1,0 Catphan 0,8 In-house MTF (lp/mm) 0,6 0,4 0,2 0,0 80 kVp 100 kVp 120 kVp Figure 4.8. Modulation transfer function measurement results from varied tube voltages on Catphan® 604 and the in-house phantom. 4.4 Novel Phantom Performance Evaluation for 3DRA 4.4.1 Comparison with Catphan® phantom The typical scan result of Catphan® phantom modules is shown in Figure 4.9. It was expected that some objects would have been identified through visual inspection on both modalities. While several objects can indeed be identified on CTA images, 3DRA images had not been able to display any object present in the phantom. The visible noise and streak pattern present on the images are also notable. Figure 4.10 presents the comparison result, indicating that SDNRs of the objects scanned using 3DRA significantly differ (p < 0.05) from those using CTA, i.e., significantly lower. As visual comparison, Figure 4.11 presents scan results of the in-house phantom using 3DRA and CTA. A quantitative report on the in-house phantom is presented in the subsequent section. 51

Figure 4.9. Scan result of Catphan®’s low contrast module using (a) 3DRA and (b) CTA. 5 p = 0.003 CTA 4 3DRA SDNR 3 2 p = 0.003 1 p = 0.008 0 0.3% 0.5% 1.0% Catphan CTP515 object Figure 4.10. Quantitative measurement (SDNR) of the Catphan® phantom scanned using 3DRA and CTA. 52

Figure 4.11. Scan result of the in-house phantom made using 3DRA and CTA. 4.4.2 In-house phantom performance for 3DRA Figure 4.12 shows the SDNRs of all simulated vessels acquired using the 3DRA system. The phantom module has managed to preserve visibility of the small object (2 mm) with the increase of SDNR due to the increase of iodine contrast agent concentration. It is of interest to observe that the SDNRs of the objects with 0.25 ml of iodine have not shown an increase with the increase of size. The trend is also shown for other concentrations when the objects are equal or larger than 6 mm. This is as expected in reconstructed 3D slices, since once the partial volume effect does not play a role anymore, only the material (or iodine concentration) in the voxels will determine the signal. Figure 4.13 shows the noise power spectrum of 3DRA images. While the integral noise magnitude for 3DRA images was measured to be 34.5 HU, it is worth notifying that the noise peaks within the range of 0.18 mm-1 to 0.26 mm- 1. These peaks range from 1500 HU2/mm2 to 2000 HU2/mm2, or around 40 HU/mm to 150 HU/mm, which intercept with the range of the HU of the contrast objects (100 HU to 250 HU) as mentioned previously. This explains the textured noise features on 3DRA images and the relatively low SDNRs. Regarding the image detail or spatial resolution, the MTF measurement result is shown in Figure 4.14. The 10% MTF was at 0.73 ± 0.01 lp/mm. 53

200 0.25 ml Iodine (a) 200 0.25 ml Iodine (b) 150 0.50 ml Iodine 0.50 ml Iodine 0.75 ml Iodine 150 0.75 ml Iodine 1.00 ml Iodine 1.00 ml Iodine SDNR SDNR 100 100 50 50 0 0 2 mm 4 mm 6 mm 8 mm 10 mm 2 mm 4 mm 6 mm 8 mm 10 mm Object size Object size Figure 4.12. Correlation between the object size and iodine concentration with the resulting SDNR for 3DRA images obtained by using the in-house phantom in bar chart (a) and line trends (b). 2500 Noise power spectra (HU2mm2) 2000 1500 fpeak: 0.27 mm-1 1000 favg: 0.34 mm-1 500 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Spatial frequency (mm-1) Figure 4.13. The noise power spectrum of 3DRA images measured from homogenous ROIs in the in-house produced phantom module. 54

Modulation Transfer Function (MTF) 1.0 3DRA MTF 10% MTF 0.8 0.6 0.4 0.2 10% MTF 0.0 0.2 0.4 0.6 0.8 1.0 1.2 0.0 Spatial frequency (lp/mm) Figure 4.14. Measurement result of MTF on in-house phantom module scanned using 3DRA. 4.4.3 In-house phantom performance for CTA Figure 4.15 shows the SDNRs of all simulated vessels acquired by using CTA. It is generally observed that SDNR grows with increasing iodine content, all with relatively clear visualization. While the measurements in the 2 mm vessel are typically lower than those in the larger vessels, there is no obvious increasing trend in the SDNR with increasing vessel size. This result indicates the potential use of the in-house phantom as a tool for image quality optimization in CTA using detectability metrics. The in-house phantom contrast objects on CTA images have typically higher SDNR (ranging from 20 to 170) when compared to 3DRA images (SDNR ranging from 10 to 70). Generally, it is observed that SDNR increases with the increase of object size, although without an obvious and linear trend. For most objects, the use of 1.0 ml iodine significantly increases the SDNR. 55

Figure 4.16 presents the measured NPS of the CTA modality measured by using the in-house phantom. With the noise peak being indicated at the frequency around 0.27 mm-1 (or visibly ranging from 0.18 mm-1 to 0.30 mm-1), the magnitude in this range is observed to be from 10 HU2mm2 to 13 HU2mm2. With the in-house phantom used for CTA, the contrast objects scanned yielded an HU range of 120 HU to 170 HU. This explicitly explains the relatively higher visibility (SDNR) of CTA (noise HU being lower than object HU) compared to 3DRA results. On the spatial resolution side, the MTF measurement result (averaged over ten measurements) is presented in Figure 4.17. The 10% MTF is 0.62 ± 0.04 lp/mm. 0.25 ml Iodine (a) 300 0.25 ml Iodine (b) 300 0.50 ml Iodine 0.50 ml Iodine 0.75 ml Iodine 0.75 ml Iodine 1.00 ml Iodine 1.00 ml Iodine 200 200 SDNR SDNR 100 100 0 0 2 mm 4 mm 6 mm 8 mm 10 mm 2 mm 4 mm 6 mm 8 mm 10 mm Object size Object size Figure 4.15. Results of SDNR measurements for CTA using the in-house phantom in bar chart (a) and line trends (b). 56

Noise power spectra (HU2mm2)14 12 Modulation Transfer Function (MTF)10 8 fpeak: 0.27 mm-1 fave: 0.32 mm-1 6 4 2 0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Spatial frequency (mm-1) Figure 4.16. The noise power spectrum of CTA measured from homogenous ROIs in the in-house produced phantom. 1.0 CTA MTF 10% MTF 0.8 0.6 0.4 0.2 10% MTF 0.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Spatial frequency (lp/mm) Figure 4.17. The 10% MTF measured by using the in-house phantom scanned with 3DRA. 57

4.4.4 Overall evaluation Phantoms are an essential tool in the process of quality assurance and one of the backbone roles for clinical medical physicists. Image quality in diagnostic and interventional radiology, and more generally, the performance of a modality, is mostly evaluated from phantoms (DeWerd & Kissick, 2014). In CT modalities, volumetric phantoms are commonly used and provided with object inserts for image quality evaluation. The same test objects may not be applicable for special purposes of CT (e.g., CTA) and CT-like three-dimensional imaging like 3DRA. This study highlights that the different natures of image acquisition in 3DRA compared to CT may lead to a need for a different phantom: while Catphan® images produced using CTA demonstrated adequate visual representation of the objects, such trend was not observed when 3DRA is used. In 3DRA, the inserts of low contrast module are not visible, probably due to the amount of noise present in 3DRA, owing to the non-full rotation of gantry during acquisition that provides lack of image projections for reconstruction. Another prominent cause is the relatively large scatter volume in 3DRA acquisition. This observation was the main reason for development and production of a new phantom. The obtained result suggests that the resulting scan images of CTA and 3DRA will present different visual representations due to the difference in image acquisition parameters. The primary difference is the use of different beam quality (5.1 mm Al in CTA and 3.0 mm Al in 3DRA), providing higher beam penetration for CTA on less scatter volume due to fan beam configuration. With less beam quality and higher scatter volume owing to the use of broader beam, the 3DRA yielded higher noise level. Additionally, the aforementioned difference in parameters has led to the difference of obtained HU ranges of the same contrast objects scanned using 3DRA and CTA; while 3DRA yielded 100 HU to 250 HU, CTA gave narrower values of 120 HU to 170 HU. This trend will be useful to explain the noise difference in subsequent evaluations. 3D imaging does not always provide true HU values. Furthermore, it can be deducted that the present tool for CT quality assurance could not suffice the need of 3DRA quality assurance task when linked to its specific clinical task. A phantom as quality control tool dedicated for 3DRA with parts or modules accommodating objects with iodine contrast equivalent is, then, considered a necessity. Meanwhile, Figure 4.18 shows typical result of the 58

in-house phantom acquired using CTA and 3DRA, with the specially designed contrast module with artificial vessels being clearly visible. Present verification of the newly designed phantom does not only provide confidence on the requirement of a new tool for quality control but can also be extended to task- based optimization. This is because the tool constructed to address this issue makes use of clinically relevant information (i.e., contrast level in vascular anatomy) that represents clinical tasks. As the clinically relevant objects can only be imaged on 3DRA through the in-house phantom, all subsequent quantitative evaluations can be performed on the in-house phantom only. Although not to be regarded as visibility threshold, the Rose model had indicated that the image of any object with SDNR > 5 against its background can be distinguished by human eye regardless of the imaging system and observer condition (Burgess, 1999; Rose, 1973), keeping in mind that objects with SDNR ≤ 5 simply may or may not be visible to human eye as they are still subjected to variations of imaging technology and observer condition. In the clinical situation, although there is no established standard of contrast enhanced vessel visibility, injection of contrast is deemed successful when vessels are visible by clinician upon fluoroscopic or fluorographic image acquisition under the constraint of contrast agent dose. Therefore, since all objects in the in-house phantom has SDNR more than 10, it can be deducted that the in-house phantom visually represents clinical situation in terms of object contrast. Additionally, the relatively higher SDNR of CTA images of the phantom is explainable by revisiting the evaluated NPS and correlating the peak noise HU with the objects HU. With this distinction against 3DRA, there might be a need to customize the iodine concentrations on the contrast module for CTA. When used for noise characterization, the in-house phantom also shows similarity in noise spectral shape as shown in Figures 4.20 (3DRA) and 4.23 (CTA), with the exception of the peaks presence in the 3DRA use which originate from the intrinsic noise character of the modality. In 3DRA use (Figure 4.20), the in-house phantom yielded the peak frequency (fpeak) of 0.27 mm-1. As for the average peak noise frequency (favg), the obtained value was 0.34 mm-1. While the integral noise magnitude for 3DRA images was measured to be 34.5 HU, noise peaked within the range of 0.18 mm-1 to 0.26 mm-1. The magnitudes of these peaks were found to be from 1500 HU2mm2 to 2000 HU2mm2, or 40 HU/mm to 150 HU/mm. This explains the relatively low SDNRs, since the noise range intercepts with the range of the HU of all contrast objects (100 HU to 250 HU). 59

The result for CTA images (Figure 4.23) denotes that although there is no demonstrable difference in the spectral shape, noise peaks are almost absent and the magnitudes are smaller, ranging from 10 HU2mm2 to 13 HU2mm2. With the in-house phantom’s contrast, the objects scanned using CTA yielded HU between 120 HU and 170 HU, the tendency of these objects to have higher SDNRs is immediately explained. Since there are limitations on the available image and the ImQuest® software had already averaged over multiple images, statistical analysis is not available in this study. Nevertheless, the level of differences in Fourier-domain noise characteristic does not lead to major spatial domain difference on the image. This implies that the in-house phantom can be proposed for evaluation of noise magnitude in CTA and 3DRA. With the in-house phantom being evaluated and deemed to be adequate for special use in 3DRA and CTA, some of its potential applications can be discussed and reserved for subsequent studies. The first proposed application is for the acceptance test and quality assurance program, considering the ability of the phantom to measure first-degree metrics of image unsharpness (i.e., MTF) and noise (NPS) with no significant difference in result compared to those measured by using Catphan®. Since the developed contrast module involving artificial vessels can be modified by adjusting iodine contrast agent concentration, the range of object contrast can also be adjusted to match the specific organ- or procedure-based clinical task. It is also a challenge to actually match the in-phantom iodine concentration to the actual in-blood iodine concentration at the imaged site for better clinical representation and sensitivity. Once such data is available, specific detectability studies using human or model observers can be carried out by using this phantom (Ba et al., 2015; Racine et al., 2016; Verdun et al., 2015). Additionally, a simple study that can be performed by medical physicists is a task-based optimization or selection of 3DRA and CTA acquisition modes with additional dose information as previously performed on planar angiography (Lubis et al., 2015; Lubis et al., 2018). An external validation of the contrast agent concentrations in the inserts should further confirm the value of the phantom. It is confirmed that the commercially available phantom dedicated for CT (Catphan®) is adequate for CT purpose while being limited for use in contrast-enhanced procedures like 3DRA and CTA, indicating that a dedicated 60

phantom for first-degree contrast and task-based image quality evaluation is required for 3DRA and CTA use. A special in-house phantom module constructed to address the need demonstrated higher object contrast visibility against Catphan®. The constructed phantom had also tested the capabilities of measuring MTF as image unsharpness metric as well as noise characterization (NPS curve). The new phantom can be proposed for multi-purpose applications, including general image quality evaluation, human- or model-observer study, as well as task-based optimization or acquisition mode selection studies. 4.5 Selection of 3DRA Task-Based Imaging Modes The aim of this final study was to select or document the optimal imaging mdoe of 3DRA of the head anatomical region. The modality has two modes intended for cranial applications with similar protocol name (‘cerebral’ and ‘head limited’ with no explanation on what the phrase ‘limited’ represent). Since the aim was to select appropriate imaging modes for head application, analysis will be focused on the use of 16 cm in-house phantom, while the no- phantom situation was used only for dose metric selection purpose (comparison of calculated and displayed KAP). 4.5.1 Dose and spectrum analysis Measurement results of KAPDFOV and displayed KAP are shown in Figure 4.18. No agreement was observed between the displayed KAP and the KAP calculated from DFOV. This is because DFOV was described as a measure of dose output with the surface of image detector being designated as measurement point to simplify the technique. It is not intended to measure the patient- attenuated dose at the detector or serve as a measure of patient dose. Since a more representative metric (e.g., AAPM TG 111 CBCT dose measurement method) requires a special type of phantom and ionization chamber which are not commonly available in diagnostic and interventional radiology departments, the KAP remains as a metric representing patient dose in this study. The resulting KAP measurement result and the X-ray mean energies of each 3DRA mode were calculated and are listed in Table 4.6. From Figure 4.18, 61

when the more clinically relevant 16 cm phantom was used, the ‘cerebral’ mode with ‘normal’ detail option gave the highest KAP, followed by the ‘head limited’ mode with ‘normal’ detail option, the ‘cerebral’ mode with ‘low’ detail option, and ‘head limited’ mode with ‘low’ detail option. The physical reason for this trend will be discussed in the following part with the X-ray spectra as primary information. 1,0 Calculated form DFOV (no phantom) Displayed (no phantom) Displayed (16 cm phantom) 0,8 KAP (Gy.cm2) 0,6 0,4 0,2 0,0 Cerebral Normal Cerebral Low Limited Normal Limited Low Head Head Selectable 3DRA modes Figure 4.18. Results of displayed KAP and KAPDFOV measurement. Figure 4.19 presents the result of normalized X-ray spectra simulation (photon fluence per mAs) of all modes when the 16 cm phantom was used, overlayed with mass attenuation coefficient curves of the Iodine contained in the contrast agent as well as ICRU-44 soft tissue, while Figure 4.20 presents the same spectrum with the tube output being taken into consideration for deeper analysis. The dose trend in Figure 4.18 can be explained by observing the normalized X-ray spectrum of each mode in Figure 4.19. The ‘cerebral’ mode with ‘normal’ detail option delivered the highest KAP to the phantom with 62

physical reasons to be discussed below. It was then followed by the ‘head limited’ mode with ‘normal’ detail option, the ‘cerebral’ mode with ‘low’ detail option, and ‘head limited’ mode with ‘low’ detail option. Table 4.6. Displayed KAP and calculated mean energy on various modes and objects used. Exposure Task-based imaging mode (protocol names and parameters modes) No phantom Tube peak Cerebral, Cerebral, Head Head voltage Tube current normal low limited, limited, low Tube output Added normal filtration KAP 65 kVp 64 kVp 80 kVp 86 kVp (Gy.cm2) Mean energy 42.8 mA 27.2 mA 1.6 mA 0.5 mA 224.7 mAs 142.8 mAs 8.4 mAs 2.6 mAs 16 cm phantom 0.3 mmCu 0.3 mmCu 0.3 mmCu 0.3 mmCu Tube peak voltage 0.413 ± 0.313 ± 0.063 ± 0.023 ± Tube current 0.006 0.006 0.006 0.005 Tube output 46.7 keV 46.3 keV 53.2 keV 55.4 keV Added filtration 65 kVp 65 kVp 75 kVp 86 kVp KAP (Gy.cm2) 54.0 mA 23.4 mA 36.4 mA 2.5 mA Mean energy 283.5 mAs 122.8 mAs 191.1 mAs 13.1 mAs 0.1 mmCu 0.1 mmCu 0.3 mmCu 0.3 mmCu 0.790 ± 0.430 ± 0.550 ± 0.120 ± 0.001 0.008 0.004 0.006 55.4 keV 42.2 keV 42.2 keV 51.2 keV Apart from the dose aspect, attention should be drawn to a very important finding from the simulated X-ray spectra results in Figure 4.19. The simulation had shown that ‘cerebral’ mode with any detail option selected had presented identical mean energy of 42.2 keV due to identical tube peak voltage and copper filter thickness. Although the mean energy is relatively closer to Iodine’s K-edge (around 33.4 keV) compared to those of the ‘head limited’ modes (51.2 keV and 63

55.4 keV for ‘normal’ and ‘low’ detail options, respectively), a considerable portion in the lower energy part of the spectrum is situated below Iodine’s K- edge, i.e., in the area where the gap between Iodine and soft tissue is considerably close. Major consequence to this is that these ‘cerebral’ modes will deliver a considerable amount of low energy photons that are deposited in the patient but do not contribute to image contrast. This is because X-rays will only be effectively absorbed by the Iodine and resulting in better image contrast when their energy is slightly higher than the peak absorbance energy of Iodine. The situation in which photon energy is lower than the absorption edge of the iodine will cause the image of the iodinated vascular to be less visible (lower contrast against its soft tissue background due to close attenuation gap), thus producing lower clinical image quality. In turn, the automatic exposure rate control (AERC) would drive the tube current into such a high level to compensate this lack of contrast, as shown in Table 4.6. This mechanism has led to the high KAP for ‘cerebral’ mode with ‘normal’ detail option as shown in Figure 4.18 and have had an impact to the SDNR as will be discussed in the next section. With this information at hand, it can be deducted that the ‘cerebral’ mode with both detail options are the modes with less appropriate beam quality to be selected for interventional procedures in which the visibility of vascular system (anatomy and/or pathology) determines the outcome of the procedure. It is more appropriate to use the ‘cerebral’ mode for non-contrast procedures, such as pre- intervention localization or other procedures focusing on anatomical landmarks other than vascular. Figure 4.19 presents that the ‘head limited’ modes, although with relatively more distant mean energy (51.2 keV and 55.4 keV for ‘normal’ and ‘low’ detail options, respectively) to Iodine’s K-edge, have almost all its photon energy distributed above Iodine’s K-edge, i.e., in the energy range where Iodine and soft tissue have a relatively large gap. This means that the ‘Head limited’ mode, with both detail selections, provides X-ray beams that are effectively useful in image contrast production while only minimally contributes to patient dose. With this information at hand, it can be deducted that the ‘head limited’ mode with the two detail options can be recommended for interventional procedures that involve iodinated contrast agent owing to the appropriateness of X-ray beam quality provided by the automated selection of tube peak voltage and copper filtration thickness. The role of the AERC has an impact to the image 64

quality outcome in terms of the SDNR owing to the compensating tube current, as will be discussed in the next section. 3e+4 Cerebral, normal 1e+4 3e+4 Cerebral, low Head limited, normal Head limited, low 2e+4Photon fluence / mAsIodine m/r 1e+3 2e+4 m/r (cm2/g)1e+2 1e+4 ICRU-44 1e+1 5e+3 soft tissue 1e+0 m/r 0 20 40 60 80 100 0 Photon energy (keV) Figure 4.19. Normalized X-ray spectra of the four 3DRA imaging modes overlayed by iodine and ICRU-44 soft tissue mass attenuation coefficient curves. 4.5.2. SDNR assessment result For all object sizes except the smallest size (2 mm), it is shown that the ‘cerebral’ mode with ‘normal’ detail option provides the highest SDNRs. However, in clinical practice, the smallest object (2 mm) will not be of particular interest since cerebral 3DRA is not commonly used for initial diagnosis and detection—rather as follow-up or inspection before or during therapeutic procedure. Moreover, any selected combination of mode and detail option can be chosen with little effect to the image contrast (SDNR). In general, the resulting SDNRs under all modes (except ‘cerebral’ mode with ‘normal’ detail option) for all objects were around 3 to 10. 65

Attention should be directed to the other object sizes (4 cm to 10 cm), since objects of this size range are more commonly found in cerebral application in the form of aneurysms. Although the ‘cerebral’ mode with ‘normal’ detail option had shown highest SDNR (and therefore highest visibility among other modes) for all object sizes, it should be noted that this is a result of high photon fluence driven by the AERC, as proven by the high KAP shown in Figure 4.18. With respect to the X-ray spectrum alone, the ‘cerebral’ mode with both detail option delivers a considerable portion of patient dose when used for procedures requiring iodinated contrast to be in place. Whereas driving up the tube current might provide some benefit in terms of SDNR, it should always be noted that lower energy photons are also increased and would contribute to higher dose. Furthermore, the ‘low’ detail option under ‘cerebral’ mode should not be selected due to relatively similar image contrast (SDNR) against its counterpart (‘head limited’ mode with ‘low’ detail option). It is of interest that the ‘head limited’ mode with ‘low’ detail option yielded a slightly higher SDNR trend when compared to the other modes except ‘cerebral’ mode with ‘normal’ detail. It is also important to note that the KAP for this mode is the lowest. This is because the mode had already given a relatively higher tube peak voltage (86 kVp) that allows the spectrum to cover the Iodine-tissue high contrast region, thus giving no need for the AERC to drive the tube current up to compensate for low contrast. Major consequence to this is the relatively higher normalized intensity than the ‘normal’ detail option. As illustrated in Figure 4.20, the ‘head limited’ mode with ‘low’ detail option had yielded on the lowest total intensity (thus lowest dose) while situated at the energy range of relatively large difference of Iodine and soft tissue absorbance—giving relatively higher SDNR than the ‘head limited’ mode under ‘normal’ detail option. This had led to the conclusion that for contrast-enhanced procedure, it is recommended to use the ‘head limited’ mode with ‘low’ detail option. A particular note should be made on the contrast agent concentration used for procedures employing ‘head limited’ mode, since the ‘head limited’ mode was deemed more appropriate for procedures using Iodinated contrast agent. Based on results, when the ‘normal’ detail option was used, the resulting SDNR would be comparable among all concentrations of Iodine. Thus, using 9.25 mg/ml will already provide adequate visibility since it is comparable with 66

those of higher concentration. On the other hand, when the ‘low’ detail option was employed, Iodine concentration of 9.25 mg/ml would present lower visibility compared to others. Therefore, a concentration of minimum 18.5 mg/ml is required to obtain the comparable SDNR with those of higher concentration when the ‘low’ detail option is selected, from which the preference follows. 5e+6 Cerebral, normal 1e+4 Cerebral, low Head limited, normal Head limited, low 4e+6 1e+3 3e+6 Photon fluenceIodine m/r m/r (cm2/g) 1e+2 2e+6 1e+1 1e+6 ICRU-44 soft tissue m/r 1e+0 0 20 40 60 80 100 0 Photon energy (keV) Figure 4.20. X-ray spectra for total photon fluence of the four 3DRA imaging modes overlayed by iodine and ICRU-44 soft tissue mass attenuation coefficient curves. 4.5.3. Qualitative analysis for mode selection With the information on KAP, beam spectrum, and image quality (SDNR) at hand, a selection of modes can be performed after a further analysis. The information on KAP was directly related with the beam spectrum and having direct impact on the image quality as basis for mode selection. The ‘cerebral’ mode with ‘normal’ detail selection had delivered the highest KAP to the 67

phantom and produced highest SDNR on the resulting image. Upon inspecting the X-ray spectrum, there is a considerable number of photons with energies below Iodine’s K-edge and in the range of close gap between Iodine and ICRU- 44 soft tissue. Theoretically, these photons were photoelectric photons with higher tendency to interact with the scanned object (phantom or patient). Since their energy is not in the range where Iodine and soft tissue largely differ in mass attenuation coefficient, they are unlikely to produce adequate contrast between the two. When uncompensated, it is very unlikely that clinicians can use this mode to visualize Iodine contrast-filled vascular system. The modality’s AERC system would then have to compensate this situation by driving up the tube current to the level that it produces sufficient number of photons to provide contrast. While this mechanism might have worked in increasing the visibility of Iodine (high SDNR), it has a detrimental effect to patient dose. Therefore, it is not recommended to use the ‘cerebral’ mode, either with ‘normal’ or ‘low’ detail selection for procedures with Iodinated contrast agents involved. This mode should be preferred for any procedures with focus other than observing vascularity with Iodine contrast agent (non-contrast procedures). Should this mode be used for such purpose, there will be no need for the AERC to compensate the low Iodine versus soft tissue contrast and patient dose would be kept low with adequate Iodine contrast-free image contrast of soft tissue. Meanwhile, the non-standard ‘head limited’ mode is more suggestible for procedures whose outcome depends on the visibility of Iodine-filled vascular system. The reason is that these modes employ relatively higher tube peak voltage (75 kVp and 86 kVp for ‘normal’ and ‘low’ detail options, respectively) and relatively thicker Copper filter (0.3 mm). This combination had shaped the X-ray spectrum to higher energy domain, conveniently in the range where Iodine and soft tissue have relatively wider mass attenuation coefficient discrepancy, but with low intensity (see Figure 4.20). That is, when imaged using this beam quality, there will be visible contrast between Iodine-filled vascular system and its background soft tissue, even without involving a high number of photons as a result of the AERC-regulated compensation. In particular, the ‘low’ detail option had served as an example where the KAP is the lowest (due to low mAs) but provided slightly higher SDNR when compared to those from ‘normal’ detail option. Therefore, it is suggested to choose the ‘head limited’ mode with ‘low’ detail options for contrast-enhanced procedures. 68

Some limitations of this study part need also to be discussed. The study used only one 3DRA-equipped angiography modality with AERC active during 3DRA acquisitions. Other known modalities from other manufacturers had static exposure parameters (AERC inactive) during 3DRA acquisition. This limits the study’s results and recommendations to the modality used in this study and other modalities with relevant AERC settings. The phantom used in this study is made of PMMA for practicability, although its attenuation coefficient differs up to 20% when compared to soft tissue. 69

CONCLUSION AND SUGGESTIONS 5.1 Conclusion The series of five studies in the dissertation had presented the required preliminary studies to arrive at a design of an in-house phantom for 3DRA performance evaluation, focused on cranial vascular applications. The produced phantom has been tested on both CT and 3DRA with results indicating propriety of use. The studies were concluded by demonstrating the application of the in- house phantom for task-based imaging mode selection for 3DRA in cranial applications. 5.2 Suggestions The importance of dedicated QC for 3DRA is often overlooked by the standard interventional fluoroscopy protocols. Therefore, series of studies should be performed to develop universal QC protocols and tools for dose and image quality. This dissertation is focused on the attempt to address this issue. However, due to the relatively novel concept, some issues still require attention. From the series of studies conducted in this dissertation, several issues are addressed as follows: 1. Further investigations are required to produce novel phantom materials that are more durable and stronger both as material insert and as phantom body. Studies in the area have been performed using carbon and resin solution, but it is still in need of technical improvement. 2. Volumetric dose descriptors for 3DRA require more thorough investigations involving measurement-validated Monte-Carlo simulations. This area of investigation is currently rare and challenging due to differences in the rotation modes among manufacturers, challenging the generalization of the new formulation. A study with Monte-Carlo simulations has been initiated but needs further improvement. 3. For modalities equipped with similar protocol names for 3DRA modes, medical physicists should be able to perform dose, image quality and 70

spectrum analysis to help clinical colleagues deciding for the best approach. To prevent confusion in the first place and to ensure proper use of each mode, effective collaboration should be in place between clinical users, medical physicists, and manufacturer’s technical representative. As a result of this, defining protocol names is crucial. The name should describe how they technically differ and highlight their effects to patient dose as well as produced image quality. 71

3BREFERENCES Attix, F. H. (1986). Introduction to Radiological Physics and Radiation Dosimetry. Wiley. Ba, A., Eckstein, M. P., Racine, D., Ott, J. G., Verdun, F., Kobbe-Schmidt, S., & Bochud, F. O. (2015). Anthropomorphic model observer performance in three-dimensional detection task for low-contrast computed tomography. Journal of Medical Imaging, 3(1), 011009. Bai, M., Liu, B., Mu, H., Liu, X., & Jiang, Y. (2012). The comparison of radiation dose between C-arm flat-detector CT (DynaCT) and multi-slice CT (MSCT): A phantom study. European Journal of Radiology, 81(11), 3577–3580. Boone, J. M., & Seibert, J. A. (1997). An accurate method for computer- generating tungsten anode x-ray spectra from 30 to 140 kV. Medical Physics, 24(11), 1661–1670. Burgess, A. E. (1999). The Rose model, revisited. Journal of the Optical Society of America A, 16(3), 633. https://doi.org/10.1364/josaa.16.000633 Bushberg, J. T., Seibert, J. A., Leidholdt, E. M., & Boone, J. M. (2011). The Essential Physics of Medical Imaging. Chu, R. Y. L., Thomas, G., & Maqbool, F. (2006). Skin entrance radiation dose in an interventional radiology procedure. Health Physics, 91(1), 41–46. de las Heras Gala, H., Torresin, A., Dasu, A., Rampado, O., Delis, H., Hernández Girón, I., Theodorakou, C., Andersson, J., Holroyd, J., Nilsson, M., Edyvean, S., Gershan, V., Hadid-Beurrier, L., Hoog, C., Delpon, G., Sancho Kolster, I., Peterlin, P., Garayoa Roca, J., Caprile, P., … Trianni, A. (2019). Quality control in cone-beam computed tomography (CBCT): EFOMP-ESTRO-IAEA protocol (2nd ed.). EFOMP-ESTRO-IAEA. de las Heras Gala, H., Torresin, A., Dasu, A., Rampado, O., Delis, H., Hernández Girón, I., Theodorakou, C., Andersson, J., Holroyd, J., Nilsson, M., Edyvean, S., Gershan, V., Hadid-Beurrier, L., Hoog, C., Delpon, G., Sancho Kolster, I., Peterlin, P., Garayoa Roca, J., Caprile, P., & Zervides, C. (2017). Quality control in cone-beam computed tomography (CBCT) EFOMP-ESTRO-IAEA protocol (summary report). Physica Medica, 39, 67–72. De Potter, T. J., Bardhaj, G., Viggiano, A., Morrice, K., & Geelen, P. (2014). 72

Three-dimensional Rotational Angiography as a Periprocedural Imaging Tool in Atrial Fibrillation Ablation. Arrhythmia & Electrophysiology Review, 3(3), 173. DeWerd, L. A., & Kissick, M. (2014). The Phantoms of Medical and Health Physics: Devices for Research and Development (L. A. DeWerd & M. Kissick (eds.)). Springer New York. Faulkner, K. (2005). Protocols for dosimetry and patient reference levels. Radiation Protection Dosimetry , 117(1–3), 195–198. https://doi.org/10.1093/rpd/nci715 Feldkamp, L. A., Davis, L. C., & Kress, J. W. (1984). Practical cone-beam algorithm. Journal of the Optical Society of America A, 1(6), 612. Friedman, S. N., Fung, G. S. K., Siewerdsen, J. H., & Tsui, B. M. W. (2013). A simple approach to measure computed tomography (CT) modulation transfer function (MTF) and noise-power spectrum (NPS) using the American College of Radiology (ACR) accreditation phantom. Medical Physics, 40(5), 051907. Gislason-Lee, A. J., McMillan, C., Cowen, A. R., & Davies, A. G. (2013). Dose optimization in cardiac x-ray imaging. Medical Physics, 40(9), 091911. Gislason, A. J., Davies, A. G., & Cowen, A. R. (2010). Dose optimization in pediatric cardiac x-ray imaging. Medical Physics, 37(10), 5258–5269. Grimmer, R., Krause, J., Karolczak, M., Lapp, R., & Kachelrie, M. (2008). Assessment of spatial resolution in CT. IEEE Nuclear Science Symposium Conference Record, 5562–5566. Haba, T., Koyama, S., Kinomura, Y., Ida, Y., & Kobayashi, M. (2017). New weighting factor of weighted CTDI equation for PMMA phantom diameter from 8 to 40 cm: A Monte Carlo study. Medical Physics, 44(12), 6603–6609. Hendee, W. R., & Ritenour, E. R. (2002). Medical Imaging Physics (4th ed.). John Wiley and Sons Ltd. Hubbel, J. H., & Seltzer, S. M. (1996). Tables of X-Ray Mass Attenuation Coefficients and Mass Energy-Absorption Coefficients from 1 keV to 20 MeV for Elements Z = 1 to 92 and 48 Additional Substances of Dosimetric Interest. International Atomic Energy Agency. (2007). IAEA Technical Report Series No. 457: Dosimetry in Diagnostic Radiology: an International Code of 73

Practice. International Atomic Energy Agency. International Atomic Energy Agency. (2014). IAEA Diagnostic Radiology Physics: A Handbook for Teachers and Students. IAEA. International Commission on Radiation Units and Measurement. (1992). ICRU Report 46: Photon, electron, proton and neutron interaction data for body tissues. In International Commission on Radiation Units & Measurements (Vol. 46). International Electrotechnical Commission. (1999). IEC 60601-2-44: Medical electrical equipment - Part 2-44: Particular requirements for the safety of X-ray equipment for computed tomography. International Electrotechnical Commission. (2009). IEC 60601-2-44: Medical electrical equipment - Part 2-44: Particular requirements for the basic safety and essential performance of X-ray equipment for computed tomography. Ishihara, S., Ross, I. B., Piotin, M., Weill, A., Aerts, H., & Moret, J. (2000). 3D Rotational Angiography: Recent Experience in the Evaluation of Cerebral Aneurysms for Treatment. Interventional Neuroradiology : Journal of Peritherapeutic Neuroradiology, Surgical Procedures and Related Neurosciences, 6(2), 85–94. Johns, H. E., & Cunningham, J. R. (1983). The Physics of Radiology. Charles C. Thomas. Judy, P. F. (1976). The line spread function and modulation transfer function of a computed tomographic scanner. Medical Physics, 3(4), 233–236. Li, M.-H., Gu, B.-X., Chen, S.-W., Wang, W., Zhang, B.-L., Li, M., Shi, W.- Y., & Li, Y.-D. (2010). 3D rotational angiography with volume rendering: The utility in the detection of intracranial aneurysms. Neurology India, 58(6), 908. Liapi, E., Hong, K., Georgiades, C. S., & Geschwind, J.-F. H. (2005). Three- dimensional Rotational Angiography: Introduction of an Adjunctive Tool for Successful Transarterial Chemoembolization. Journal of Vascular and Interventional Radiology, 16(9), 1241–1245. Loveland, J. (2011). SPICE-CT [computer software]. Lubis, L. E., Bayuadi, I., Pawiro, S. A., Ng, K.-H., Bosmans, H., & Soejoko, D. S. (2015). Optimization of dose and image quality of paediatric cardiac catheterization procedure. Physica Medica, 31(7), 659–668. Lubis, L. E., Craig, L. A., Bosmans, H., & Soejoko, D. S. (2018). Task-based 74

phantom evaluation of cardiac catheterization imaging modes. Physica Medica, 46, 114–123. Markovich, A., Morgan, A. G., Dong, F. F., Primak, A. N., & Li, X. (2017). Extending the concept of weighted CT dose index to elliptical phantoms of various aspect ratios. Journal of Medical Imaging (Bellingham, Wash.), 4(3), 031205. McFadden, S. L., Mooney, R. B., & Shepherd, P. H. (2002). X-ray dose and associated risks from radiofrequency catheter ablation procedures. The British Journal of Radiology, 75(891), 253–265. Mieville, F., Beaumont, S., Torfeh, T., Gudinchet, F., & Verdun, F. R. (2010). Computed tomography commissioning programmes: how to obtain a reliable MTF with an automatic approach? Radiation Protection Dosimetry, 139(1–3), 443–448. Minerbo, G. N. (1979). Convolutional reconstruction from cone-beam projection data. IEEE Transactions on Nuclear Science, 26(2), 2682– 2684. Möhler, C., Wohlfahrt, P., Richter, C., & Greilich, S. (2017). Methodological accuracy of image-based electron density assessment using dual-energy computed tomography. Medical Physics, 44(6), 2429–2437. Nakaya, Y., Kawata, Y., Niki, N., Umetatni, K., Ohmatsu, H., & Moriyama, N. (2012). A method for determining the modulation transfer function from thick microwire profiles measured with x-ray microcomputed tomography. Medical Physics, 39(7), 4347–4364. Nalcioglu, O., & Cho, Z. H. (1978). Reconstruction of 3-D Objects from Cone Beam Projections. Proceedings of the IEEE, 66(11), 1584–1585. Neubauer, A. M., Garcia, J. A., Messenger, J. C., Hansis, E., Kim, M. S., Klein, A. J. P., Schoonenberg, G. A. F., Grass, M., & Carroll, J. D. (2010). Clinical Feasibility of a Fully Automated 3D Reconstruction of Rotational Coronary X-Ray Angiograms. Circulation: Cardiovascular Interventions, 3(1), 71 LP – 79. Nollert, G., Hartkens, T., Figel, A., Bulitta, C., Altenbeck, F., & Gerhar, V. (2012). The Hybrid Operating Room. In Special Topics in Cardiac Surgery. InTech. Racine, D., Ba, A. H., Ott, J. G., Bochud, F. O., & Verdun, F. R. (2016). Objective assessment of low contrast detectability in computed tomography with Channelized Hotelling Observer. Physica Medica, 75

32(1), 76–83. Rasche, V., Buecker, A., Grass, M., Koppe, R., Op de Beek, J., Bertrams, R., Suurmond, R., Kuehl, H., & Guenther, R. W. (2002). ECG-gated 3D- rotational coronary angiography (3DRCA). In CARS 2002 Computer Assisted Radiology and Surgery (pp. 827–831). Springer Berlin Heidelberg. Ria, F., Solomon, J. B., Wilson, J. M., & Samei, E. (2020). Technical Note: Validation of TG 233 phantom methodology to characterize noise and dose in patient CT data. Medical Physics, mp.14089. Richard, S., Husarik, D. B., Yadava, G., Murphy, S. N., & Samei, E. (2012a). Towards task-based assessment of CT performance: System and object MTF across different reconstruction algorithms. Medical Physics, 39(7), 4115–4122. Richard, S., Husarik, D. B., Yadava, G., Murphy, S. N., & Samei, E. (2012b). Towards task-based assessment of CT performance: System and object MTF across different reconstruction algorithms. Medical Physics, 39(7Part1), 4115–4122. Rose, A. (1973). Vision: Human and Electronic. Plenum Press. Rossmann, K. (1969). Point Spread-Function, Line Spread-Function, and Modulation Transfer Function. Radiology, 93(2), 257–272. Saito, M., & Sagara, S. (2017). A simple formulation for deriving effective atomic numbers via electron density calibration from dual-energy CT data in the human body. Medical Physics, 44(6), 2293–2303. Shilfa, S. N., Gani, M. R. A., Mu’minah, I. A. S., Ardiansyah, F., Lubis, L. E., & Soejoko, D. S. (2019). Pengukuran MTF (Modulation Transfer Function) berdasarkan LSF (Line Spread Function) dan PSF (Point Spread Function) pada pesawat PET/CT dan SPECT/CT. Journal of Medical Physics and Biophysics, 6(1), 16–25. Shope, T. B., Gagne, R. M., & Johnson, G. C. (1981). A method for describing the doses delivered by transmission x-ray computed tomography. Medical Physics, 8(4), 488–495. Solomon, J., Zhang, Y., Wilson, J., & Samei, E. (2018). An automated software tool for task-based image quality assessment and matching in clinical CT using the TG-233 Framework. Medical Physics, 45(6), E134. Stenger, A., Dittrich, S., & Glöckler, M. (2016). Three-Dimensional Rotational Angiography in the Pediatric Cath Lab: Optimizing Aortic Interventions. 76

Pediatric Cardiology, 37(3), 528–536. Strauss, K. J. (2002). Interventional Equipment Acquisition Process: Cradle to Grave. In S. Balter, R. C. Chan, & T. B. Shope Jr. (Eds.), Medical Physics Monograph No. 28: Intravascular Brachytherapy, Fluoroscopically Guided Interventions. Medical Physics Publishing. Tacher, V., Radaelli, A., Lin, M., & Geschwind, J.-F. (2015). How I do it: Cone-beam CT during transarterial chemoembolization for liver cancer. Radiology, 274(2), 320–334. Takenaga, T., Katsuragawa, S., Goto, M., Hatemura, M., Uchiyama, Y., & Shiraishi, J. (2015). Modulation transfer function measurement of CT images by use of a circular edge method with a logistic curve-fitting technique. Radiological Physics and Technology, 8(1), 53–59. The International Commission on Radiation Unit and Measurement. (2005). ICRU Report 74: Patient Dosimetry for X Rays Used in Medical Imaging. In Journal of the ICRU (Vol. 5, Issue 2). Oxford University Press. Tudor, G. S. J., & Thomas, S. J. (2004). Equivalent diameters of elliptical fields. The British Journal of Radiology, 77(923), 941–943. van Rooij, W. J., Sprengers, M. E., de Gast, A. N., Peluso, J. P. P., & Sluzewski, M. (2008). 3D Rotational Angiography: The New Gold Standard in the Detection of Additional Intracranial Aneurysms. American Journal of Neuroradiology, 29(5), 976 LP – 979. Vano, E., Geiger, B., Schreiner, A., Back, C., & Beissel, J. (2005). Dynamic flat panel detector versus image intensifier in cardiac imaging: dose and image quality. Physics in Medicine and Biology, 50(23), 5731–5742. Vano, E., Gonzalez, L., Guibelalde, E., Aviles, P., Fernandez, J. M., Prieto, C., & Galvan, C. (2005). Evaluation of risk of deterministic effects in fluoroscopically guided procedures. Radiation Protection Dosimetry , 117(1–3), 190–194. Verdun, F. R., Racine, D., Ott, J. G., Tapiovaara, M. J., Toroi, P., Bochud, F. O., Veldkamp, W. J. H., Schegerer, A., Bouwman, R. W., Giron, I. H., Marshall, N. W., & Edyvean, S. (2015). Image quality in CT: From physical measurements to model observers. Physica Medica, 31(8), 823– 843. Yohannes, I., Kolditz, D., & Kalender, W. A. (2011). Semiempirical analysis of materials’ elemental composition to formulate tissue-equivalent 77

materials: a preliminary study. Physics in Medicine and Biology, 56(10), 2963–2977. Zhang, L.-J., Wu, S.-Y., Niu, J.-B., Zhang, Z.-L., Wang, H. Z., Zhao, Y.-E., Chai, X., Zhou, C.-S., & Lu, G.-M. (2010). Dual-Energy CT Angiography in the Evaluation of Intracranial Aneurysms: Image Quality, Radiation Dose, and Comparison With 3D Rotational Digital Subtraction Angiography. American Journal of Roentgenology, 194(1), 23–30. 78

PUBLICATION LIST By the time this dissertation is examined, some parts of the study have reached a different stage of publication as follows; • The first preliminary study is under review in Journal of Radiation Physics and Chemistry. • The second preliminary study was presented as poster and oral presentation in 2019 South East Asian Congress in Medical Physics in Bali, Indonesia. • The first primary study has been published in Atom Indonesia (vol. 46, no. 2, pp. 69-75, 2020). • The second primary study has been published in Physica Medica: European Journal of Medical Physics (vol. 90, pp. 91-98, 2021). 79

AUTHOR INFORMATION PERSONAL INFORMATION Full name : Lukmanda Evan Lubis Sex : Male Birthplace/date : Jakarta, January 15, 1989 Work address : Department of Physics, FMIPA Universitas Indonesia, Kampus UI Depok, 16424, Jawa Barat, Indonesia Home address : Jl. Kiray No. 28, Pangkalan Jati Baru, Depok, 16513, Jawa Barat Phone : 021 727 0160 / 085717381069 E-mail : [email protected] FORMAL EDUCATION Period Institution Title awarded Doctor (Dr.) 2018-2021 Universitas Indonesia Magister Sains (M.Sc.) Sarjana Sains (B.Sc.) 2012-2014 Universitas Indonesia 2007-2012 Universitas Indonesia PROFESSIONAL INFORMATION Period Position Institution UI Hospital May 2021-present Medical physicist Center for Medical Physics and June 2018-present Manager for Education Biophysics (CMPB), FMIPA UI and Training programs Dept. of Physics, FMIPA UI October 2016-present Lecturer Medphys Laboratory, Dept. of Physics, FMIPA June 2015-present Supervisor UI Medphys Laboratory, July 2013-June 2015 Qualified tester Dept. of Physics, FMIPA UI 80

ORGANIZATIONAL AFFILIATIONS Period Position Organization 2019-present Secretary General Indonesian Association of Physicists in Medicine (Aliansi Fisikawan Medik Indonesia, AFISMI) 2015-2019 Deputy Secretary General Indonesian Association of Physicists in Medicine (Aliansi Fisikawan Medik Indonesia, AFISMI) 2014-present International Affiliate American Association of Physicists in Medicine (AAPM) SCIENTIFIC PUBLICATIONS Lubis L.E., Basith R.A., Hariyati I., Ryangga D., Mart T., Bosmans H., Soejoko, D.S. Novel phantom for performance evaluation of contrast- enhanced 3D rotational angiography. Phys Medica 2021;90:91–8. https://doi.org/10.1016/J.EJMP.2021.09.002. Delis, H., Homolka, P., Chapple, C.L., Costa, P.R., Attalla, E., Lubis, L.E., Sackey, T.A., Fahey, F., Lassmann, M., Poli, G.L., 2021. Developing and implementing a multi-modality imaging optimization study in paediatric radiology: Experience and recommendations from an IAEA coordinated research project. Phys. Medica 82, 255–265. https://doi.org/10.1016/j.ejmp.2021.02.009 Lubis, L.E., Cokrokusumo, H.B., Basith, R.A., Lestariningsih, I., Prajitno, P., Soejoko, D.S., 2021. Noise reduction of three-dimensional rotational angiography (3DRA) images using residual encoder-decoder convolutional neural network (RED-CNN), in: AIP Conference Proceedings. AIP Publishing LLC AIP Publishing, p. 040007. https://doi.org/10.1063/5.0048077 Lubis, L.E., Hariyati, I., Ryangga, D., Mu’minah, I.A.S., Mart, T., Soejoko, 81

D.S., 2020. Construction and Evaluation of a Multipurpose Performance Check Phantom for Computed Tomography. Atom Indones. 46, 69. https://doi.org/10.17146/aij.2020.1004 Lubis, L.E., Soejoko, D.S., 2020. Optimisasi dosis dan kualitas citra pada radiologi diagnostik: langkah-langkah, tips, dan panduan praktis. J. Med. Phys. Biophys. 7, 22–31. Ramadhan, M.M., Faza, A., Lubis, L.E., Yunus, R.E., Salamah, T., Handayani, D., Lestariningsih, I., Resa, A., Alam, C.R., Prajitno, P., Pawiro, S.A., Sidipratomo, P., Soejoko, D.S., 2020. Fast and accurate detection of Covid-19-related pneumonia from chest X-ray images with novel deep learning model. Apriliastri, N.N., Samiyah, Bawono, S., Susilo, A., Lubis, L.E., Evianti, A., Soejoko, D.S., 2019. Optimization of simulated cranial, thorax, and abdominal examination in paediatric digital radiography. J. Phys. Conf. Ser. 1248, 012023. https://doi.org/10.1088/1742-6596/1248/1/012023 Ardyanti, E.A., Gani, M.R.A., Lubis, L.E., Soejoko, D.S., 2019. Pengaruh Antiscatter grid Terhadap Dosis dan Kualitas Citra pada Prosedur Radiologi Intervensional. J. Med. Phys. Biophys. 6, 7–15. Hariyati, I., Hani, A.D.F., Craig, L.A., Lestariningsih, I., Lubis, L.E., Soejoko, D.S., 2019. Optimization of digital radiography system using in-house phantom: preliminary study. J. Phys. Conf. Ser. 1248, 012021. https://doi.org/10.1088/1742-6596/1248/1/012021 Lestariningsih, I., Lubis, L.E., Nurlely, Soejoko, D.S., 2019. Effect of pitch on CT image quality based on SNR evaluation using in house phantom. J. Phys. Conf. Ser. 1248, 012027. https://doi.org/10.1088/1742- 6596/1248/1/012027 Lubis, L.E., Jundi, A.F., Susilo, A., Evianti, A., Soejoko, D.S., 2019. Local dose survey on paediatric multi-detector CT: A preliminary result, in: IFMBE Proceedings. https://doi.org/10.1007/978-981-10-9023-3_105 Pawiro S. A. and Lubis, L.E. and O.A.N. and S.D.S., 2019. Overseeing the Growth of Medical Physics: Indonesia Case, in: Lhotska Lenka and Sukupova, L. and L.I. and I.G.S. (Ed.), World Congress on Medical Physics and Biomedical Engineering 2018. Springer Singapore, Singapore, pp. 859–863. Pawiro, S.A., Lubis, L.E., Oktavianto, A.N., Soejoko, D.S., 2019. Overseeing the growth of medical physics: Indonesia case, in: IFMBE Proceedings. 82

https://doi.org/10.1007/978-981-10-9035-6_159 Rahman, I.N.F., Fajar, M.I., Aini, N., Febrianti, R.H., Lestariningsih, I., Gani, M.R.A., Lubis, L.E., Soejoko, D.S., 2019. Using in-house quick-QC phantom to characterize computed and direct digital radiography: a preliminary study. J. Phys. Conf. Ser. 1248, 012024. https://doi.org/10.1088/1742-6596/1248/1/012024 Shilfa, S.N., Gani, M.R.A., Mu’minah, I.A.S., Ardiansyah, F., Lubis, L.E., Soejoko, D.S., 2019. Pengukuran MTF (Modulation Transfer Function) berdasarkan LSF (Line Spread Function) dan PSF (Point Spread Function) pada pesawat PET/CT dan SPECT/CT. J. Med. Phys. Biophys. 6, 16–25. Harfah, H., Lubis, L.E., Wigati, K.T., Soejoko, D.S., 2018. Metode line profile: pendekatan terhadap evaluasi kuantitatif citra Computed Radiography thoraks pada pasien pediatrik. J. Med. Phys. Biophys. 5, 155–172. Lubis, L.E., Bayuadi, I., Bayhaqi, Y.A., Ardiansyah, F., Setiadi, A.R., Sugandi, R.D., Craig, L.A., Nasir, A., Basith, R.A., Pawiro, S.A., Soejoko, D.S., 2018. RADIATION DOSE FROM DENTAL RADIOGRAPHY IN INDONESIA: A FIVE-YEAR SURVEY. Radiat. Prot. Dosimetry. https://doi.org/10.1093/rpd/ncy123 Lubis, L.E., Craig, L.A., Bosmans, H., Soejoko, D.S., 2018. Task-based phantom evaluation of cardiac catheterization imaging modes. Phys. Medica 46, 114–123. https://doi.org/10.1016/j.ejmp.2018.02.002 Yuliani, S., Lubis, L.E., Nurlely, N., Soejoko, D.S., 2018. Kuantisasi dan analisis citra computed radiography pada pemeriksaan sinus paranasal pasien pediatrik dengan metode line profile. J. Med. Phys. Biophys. 5, 139–154. Sari, N.L.K., Prajitno, P., Lubis, L.E., Soejoko, D.S., 2017. Computer Aided Diagnosis (CAD) for mammography with Markov Random Field method with Simulated Annealing optimization. J. Med. Phys. Biophys. 4, 85–94. Lubis, L.E., Badawy, M.K., 2016. Measuring radiation dose to patients undergoing fluoroscopically-guided interventions. J. Phys. Conf. Ser. 694, 012049. https://doi.org/10.1088/1742-6596/694/1/012049 Mubarok, S., Lubis, L.E., Pawiro, S.A., 2016. Parameter-based estimation of CT dose index and image quality using an in-house androidTM-based software. J. Phys. Conf. Ser. 694, 012037. https://doi.org/10.1088/1742- 6596/694/1/012037 Purwaningsih, S., Lubis, L.E., Pawiro, S.A., Soejoko, D.S., 2016. Measurement 83

of computed tomography dose profile with pitch variation using Gafchromic XR-QA2 and thermoluminescence dosimeter (TLD). J. Phys. Conf. Ser. 694, 012046. https://doi.org/10.1088/1742-6596/694/1/012046 Lubis, L.E., Aida, N., Pratiwi, N.G., Pawiro, S.A., Wigati, K.T., Soejoko, D.S., 2015. Occupational Dose Measurement in an Interventional Radiology Facility in Jakarta, in: Jaffray, D.A. (Ed.), World Congress on Medical Physics and Biomedical Engineering, June 7-12, 2015, Toronto, Canada. Springer International Publishing, pp. 775–777. https://doi.org/10.1007/978-3-319-19387-8_191 Lubis, L.E., Bayuadi, I., Pawiro, S.A., Ng, K.-H., Bosmans, H., Soejoko, D.S., 2015. Optimization of dose and image quality of paediatric cardiac catheterization procedure. Phys. Medica 31, 659–68. https://doi.org/10.1016/j.ejmp.2015.05.011 Round, W.H., Jafari, S., Kron, T., Azhari, H.A., Chhom, S., Hu, Y.M., Mauldon, G.F., Cheung, K.Y., Kuppusamy, T., Pawiro, S.A., Lubis, L.E., Soejoko, D.S., Haryanto, F., Endo, M., Han, Y., Suh, T.S., Ng, K.H., Luvsan-Ish, A., Maung, S.O., Chaurasia, P.P., Jafri, M.A., Farrukh, S., Peralta, A., Toh, H.J., Shiau, A.C., Krisanachinda, A., Suriyapee, S., Vinijsorn, S., Nguyen, T.C., 2015. Brief histories of medical physics in Asia-Oceania. Australas. Phys. Eng. Sci. Med. https://doi.org/10.1007/s13246-015-0342-9 Round, W.H., Jafari, S., Kron, T., Azhari, H.A., Chhom, S., Hu, Y., Mauldon, G.F., Cheung, K.Y., Kuppusamy, T., Pawiro, S.A., Lubis, L.E., Soejoko, D.S., Haryanto, F., Endo, M., Han, Y., Suh, T.S., Ng, K.H., Luvsan-Ish, A., Maung, S.O., Chaurasia, P.P., Jafri, S.M.A., Farrukh, S., Peralta, A., Toh, H.J., Sarasanandarajah, S., Shiau, A.C., Krisanachinda, A., Suriyapee, S., Vinijsorn, S., Nguyen, T.C., 2015. Erratum to: Brief histories of medical physics in Asia-Oceania [Australas Phys Eng Sci Med DOI 10.1007/s13246-015-0342-9]. Australas. Phys. Eng. Sci. Med. 38. https://doi.org/10.1007/s13246-015-0370-5 Soejoko, D.S., Pawiro, S.A., Lubis, L.E., 2015. Medical Physics in Indonesia: Current Status and Plans, in: Jaffray, D.A. (Ed.), World Congress on Medical Physics and Biomedical Engineering, June 7-12, 2015, Toronto, Canada. Springer International Publishing, pp. 1601–1603. https://doi.org/10.1007/978-3-319-19387-8_389 Lubis, L.E., Bayuadi, I., Pawiro, S.A., Soejoko, D.S., 2014. Optimization of 84

Paediatric Cardiac Catheterization Using Figure of Merit (FOM) as Parameter, in: 14th Asia-Oceania Congress of Medical Physics (AOCMP) and 12th South-East Asian Congress of Medical Physics (SEACOMP). Ho Chi Minh. Bayhaqi, Y.A., Lubis, L.E., Susila, I.P., Soejoko, D.S., 2012. Volumetric reconstruction from CT images for treatment planning of external radiotherapy, in: Abstracts of the 2011 South East Asian Congress of Medical Physics (SEACOMP 2011). Springer Berlin Heidelberg, pp. 365– 387. https://doi.org/10.1007/s13246-012-0154-0 Lubis, L.E., Soejoko, D.S., 2012. Comparison between Fermi-Eyges (Hogstrom Model) Algorithm Calculation and Measurement on Percentage Depth Dose of Electron Beam, in: 12th Asia-Oceania Congress of Medical Physics (AOCMP) and 10th South-East Asian Congress of Medical Physics (SEACOMP). Chiang Mai, pp. 245–248. Lubis, L.E., Taruno, W.P., Soejoko, D.S., 2012. Effects of electric field AC frequency on contact dermatitis lesion regression: a preliminary study, in: Abstracts of the 2011 South East Asian Congress of Medical Physics (SEACOMP 2011). pp. 365–387. https://doi.org/10.1007/s13246-012- 0154-0 Soejoko, D.S., Lubis, L.E., 2012. How to Enhance Development of Medical Physics in Indonesia, in: Engineering and Physical Sciences in Medicine Conference. Gold Coast. 85