RPS 913 - Spectral imaging: quantification and performance evaluation

Author Block: L. Lofino, F. Schwartz, F. Ria, M. Z. Zarei, E. Samei, A. Abadia, D. Marin; Durham, NC/US
Purpose: The purpose of this study was to compare the image quality of portal venous phase (PVP) abdominal CT examinations and virtual non-contrast images (VNC) between photon-counting CT (PCCT) and energy-integrating Detector CT (EID).
Methods or Background: Multi-phase CT scans from one PCCT and two EID CTs were retrieved. 45 BMI-matched patients were included: 15 for PCCT and 30 for EID. In vivo image quality parameters were measured and compared for PVP and VNC. CTDIvol values were also recorded for all examinations. Because scanner tube current modulation adapts to patient size, the radiation dose was compared among scanners accounting for BMI using a figure of merit: FOM=1/(BMI*lnCTDIvol). A five-point scale was used to assess the reader's perception of image quality.
Results or Findings: Compared to the two EID, PCCT yielded significantly improved resolution and noise magnitude for both PVP (MTFf10 = 0.55 ± 0.08 for PCCT vs. 0.50 ± 0.04 and 0.49 ± 0.03 for Flash and Force; noise = 9.76 ± 3.10 vs. 15.35 ± 4.14 and 10.70 ± 1.34) and VNC (MTFf10 = 0.56 ± 0.01 for PCCT vs. 0.51 ± 0.05 and 0.51 ± 0.03 for Flash and Force; noise = 9.59 ± 2.77 vs. 13.90 ± 3.57 and 10.83 ± 2.83), P<0.02. A similar trend was confirmed in a subset of overweight patients. Our FOM analysis suggests that, for equal radiation exposure levels and comparable patient size, PCCT yields 20% noise reduction compared to the two EID, with 18% reduction in overweight patients. Reader’s perceived image noise was lower and overall image quality was higher for PCCT compared to EID.
Conclusion: PCCT yields a significantly lower radiation dose, with improved image quality in both the PVP and VNC of abdominal CT examinations.
Limitations: This was a single center retrospective study with a limited number of cases.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Yes
Ethics committee - additional information: An ethics committee approved the study.

Author Block: B. Tariq¹, N. Werghi¹, E. I. Memisoglu¹, N. Maalej¹, O. Sikander², F. Naseer², A. Y. Y. Raja¹; ¹Abu Dhabi/AE, ²Islamabad/PK
Purpose: Metal artefacts in computed tomography (CT) obscure the visualisation and assessment of anatomical structures and lead to misinterpretation of patient diagnosis and treatment. This study aims to use photon-counting spectral CT and evaluate the identification and quantification of clinically relevant materials such as iodine (as a contrast agent) and hydroxyapatite (a mineral in bones and teeth) in the presence of metal objects.
Methods or Background: A multi-material phantom was used with inserts of varied quantities of iodine (4.83, 9.66, and 14.56 mg/cm3) and hydroxyapatite (201.6 and 406.9 mg/cm^3). Three sets of scans were acquired: one without a metal insert, one with steel, and one with aluminium. Image acquisition used a Mars spectral scanner (Microlab 5x120) operated at 118 kVp, 80 μA, and 160 ms with 981 circular projections. Images were reconstructed in five energy bins: 7-40, 40-50, 50-60, 60-79, and 79-118 keV. Energy and material-density images were assessed by linear regression, sensitivity, specificity, area under the curve (AUC), and root-mean-square-error (RMSE). For demonstrative purposes, a biological sample (a sheep heart) with a steel insert was also scanned and evaluated.
Results or Findings: Results indicate reduced metal artefacts and an enhanced signal-to-noise ratio (up to 25%) in the higher energy bins. All energy bins revealed strong linearity (R^2>0.97) across the concentrations of material. Material identification and quantification were measured for iodine (without metal- Sensitivity 80%; Specificity 90%; AUC 0.80; RMSE 22%; with metal- Sensitivity >77%; Specificity >90%; AUC >0.74; RMSE 26%) and hydroxyapatite (without metal- Sensitivty 83%; Specificity 93%; AUC 0.8; RMSE,14%; with metal- Sensitivity >78%; Specificity >93%; AUC >0.77; RMSE >22%). Images of the biological sample showed comparable results to the multimaterial phantom.
Conclusion: SPCCT can accurately identify and quantify clinically relevant materials, such as iodine and hydroxyapatite, in the presence of metal objects.
Limitations: No limitations were identified.
Funding for this study: Funding was received through the Research and Innovation grant from the Khalifa University, UAE (Project#: 8474000563).
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: No information provided by the submitter.

Author Block: S. Blazis¹, T. Ruytenberg², N. De Jong², J. Heemskerk²; ¹Goes/NL, ²Leiden/NL
Purpose: The purpose of this study was to determine (1) whether quantification of iodinated contrast medium of a kVp-switching dual-energy CT is accurate, and (2) what the limit of this quantification is.
Methods or Background: Dual-energy CT has been proposed as a tool for semi-quantitative perfusion imaging, using iodine concentration maps. We have investigated the accuracy and limitations of iodine quantification of a fast kVp-switching DECT. Two phantoms were scanned: a Gammex multi-energy phantom, and a 180mm diameter cylindrical water-filled (‘linearity’) phantom containing twelve 16mm diameter tubes with iodinated contrast medium with concentrations ranging from 0 to 30 mgI/ml.
Results or Findings: In line with what has been documented elsewhere, measurements with the ‘linearity’ phantom indicate that lower iodine concentrations (<2mgI/ml) are underestimated with 60-20%. For higher concentrations, the accuracy of iodine quantification is in the order of a few percent. Nevertheless, depicted iodine concentrations increase monotonously with actual concentrations, indicating potential for scoring purposes or thresholding for clinical applications. Gammex phantom images show that distinction between iodine and calcium is challenging for the kVp-switching DECT. This is also apparent from VNC images, where iodine concentrations >3mgI/ml are no longer fully suppressed, and HU values for calcium inserts are reduced.
Conclusion: Dual-energy CT is increasingly proving itself as a useful expansion of CT imaging. Discrimination and quantification of iodinated contrast medium is a promising additional functionality for, for example, detection of intra-cranial hemorrhage or myocardial defect. Fast kVp-switching DECT switching facilitates this iodine quantification, but within limits, and allows creation of virtual non-contrast maps at no extra dose cost. Quantitative iodine imaging with CT remains challenging for clinically relevant concentrations <3mgI/ml, and further investigation of accuracy of iodine quantification and VNC-maps is warranted.
Limitations: This project included phantoms scans only.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: This project only included phantom scans.

Author Block: P. Liu, S. Zhou, Z. Xu, H. Dong, J. Li, S. Lin, W. Yang, F. Yan, L. Qin; Shanghai/CN
Purpose: The objective of this study was to investigate the feasibility and accuracy of iodine quantification and monoenergetic attenuation using PCD-CT in standard pitch and high-pitch scanning at different scan parameters in a phantom model.
Methods or Background: Four inserts with known iodine concentrations (2, 5, 10, and 15 mg/mL) were placed in the removable CT phantom and scanned using high-pitch (3.2) and standard pitch (0.8) on PCD-CT. Two tube voltages (120 and 140 kVp) and four radiation doses (1, 3, 5, and 10 mGy) were alternated. Each scan setting was repeated three times. Mean iodine concentration and monoenergetic attenuation were recorded. Percentage absolute bias (PAB) was assessed for iodine quantification. Image noise and monoenergetic attenuation were evaluated at 40, 70, 100, and 140 keV.
Results or Findings: 96 acquisitions were conducted. In small phantom, the PAB was 2.96% (1.75%, 4.56%) and 1.67% (1.00%, 3.42%) for high-pitch and standard pitch, respectively. In large phantom, these numbers were 3.72% (1.75%, 5.97%) and 2.94% (1.75%, 4.70%) respectively. Linear regression analysis revealed that only phantom size significantly influenced (P < 0.001) the accuracy of iodine quantification. Background noise increased with a decrease in keV level and radiation doses. Attenuation errors at 70, 100, and 140 keV remained below 10 HU, with 37.5% cases surpassing 10 HU at 40 keV. Linear regression analysis revealed comparable accuracy of monoenergetic attenuation between high-pitch and standard pitch (P=0.332).
Conclusion: High-pitch scanning in PCD-CT can be used to quantify iodine density and monoenergetic CT values with high accuracy, thereby potentially benefiting multienergy-based tissue differentiation and material decomposition in clinical settings.
Limitations: The results of our study were focused on a phantom and have not been verified in patients, which must be conducted in the future.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: The study is a phantom experiment.

Author Block: T. W. Holmes¹, N. Nezami², A. Jobayer³, R. Bujila⁴, J. Maltz³, Z. Yin⁵, A. Pourmorteza¹; ¹Atlanta, GA/US, ²Baltimore, MD/US, ³Waukesha, WI/US, ⁴Stockholm/SE, ⁵Niskayuna, NY/US
Purpose: Photon-counting detectors (PCD) have helped improve diagnostic value of CT by providing spectral information, increasing the spatial resolution, and reducing the radiation dose. The combination of these advantages could enable dose-efficient ultra-high-resolution (UHR) multi-contrast imaging by using the K-edges of materials. Properly timed injections of two contrast agents can generate co-registered maps of the anatomy in different perfusion phases and eliminate multi-phasic examinations. Moreover, use of solid k-edge materials such as gold and tungsten as fiducial markers in surgical tools could improve CT-guided interventions. Si-based PCDs with eight adjustable thresholds are a new development that promise better spatial and spectral resolution compared to the existing CdTe/CZT technologies. We investigated the utility of Si-PCD in multi-contrast UHR tasks.
Methods or Background: We prepared a series of anthropomorphic phantoms mimicking coronary arteries with atherosclerotic plaques and stents, tumours embolised with radiopaque microspheres, and human head including various concentrations and mixtures of iodine- and gadolinium-based contrast agents. Another series were made including different components of a 1-mm micro-robot designed for neurosurgery containing solid neodymium and tungsten components. The phantoms were scanned on a whole-body prototype Si-PCD scanner and material-specific UHR images were generated. We compared iodine and gadolinium quantification, image resolution, and material separation of the system to a state-of-the-art dual-energy energy-integrating detector (EID) CT scanner.
Results or Findings: While the dual-energy EID system failed at separating the two materials, 8-energy Si-PCD could distinguish them. Iodine quantification accuracy was not significantly different between the two systems in the absence of gadolinium. Gadolinium accuracy was [-0.3 0.7] mgGd/mL for Si-PCD. Spatial resolution was significantly higher in Si-PCD as measured by in-stent and free lumen diameters.
Conclusion: Si-PCD has the potential to improve spatial resolution and provide multi-material quantification simultaneously.
Limitations: This study utilised a prototype scanner.
Funding for this study: This study was part of a sponsored research agreement with GE Healthcare.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: No information provided by the submitter.

Author Block: S. Tilley, J. Potipcoe, K. S. Karim; Waterloo, ON/CA
Purpose: The purpose of this study was to investigate the feasibility of dual energy (DE) tomosynthesis using a triple-layer x-ray detector. Of particular interest is whether a set of DE images, acquired with a tomosynthesis acquisition protocol and processed using a traditional DE algorithm, can be reconstructed into material specific three-dimensional volumes. Adding the material separation capabilities of DE with the depth localisation of tomosynthesis may provide enough patient information to negate the costly use of computed tomography (CT) for many patients. Furthermore, using a multi-layer detector for DE as opposed to multiple exposures may permit portable, spectral tomosynthesis.
Methods or Background: We acquired 61 projection images of a chest phantom using the Reveal 35C x-ray detector. The x-ray source was linearly translated parallel to the phantom’s longitudinal axis to produce an angular sweep of 30 degrees. A a standard digital radiograph (DR) and two DE images, bone and soft-tissue, were calculated from each triple-layer projection image. A variant of logarithmic subtraction was used for the DE images, using the same parameters for each projection. We reconstructed these three datasets using the same model-based conjugate gradient algorithm with quadratic regularisation and a 5mm slice thickness.
Results or Findings: Three sets of tomographic slice data were reconstructed: DR, bone, and soft-tissue. Structures in these images were successfully separated by slice (localisation) and DE image (material identification). The DE data used the same reconstruction algorithm as the DR data, indicating that tomosynthesis can be readily applied to DE images.
Conclusion: This proof-of-concept study demonstrates the feasibility of DE tomosynthesis using a triple-layer detector. Future work will explore dose requirements, sensitivity to various pathologies, improved reconstruction techniques, and application in a portable system.
Limitations: No limitations were identified.
Funding for this study: No funding was received for this study.
Has your study been approved by an ethics committee? Not applicable
Ethics committee - additional information: No information provided by the submitter.