Optimal conspicuity of pancreatic ductal adenocarcinoma in virtual monochromatic imaging reconstructions on a photon-counting detector CT: comparison to conventional MDCT

Purpose To analyze the conspicuity of pancreatic ductal adenocarcinoma (PDAC) in virtual monoenergetic images (VMI) on a novel photon-counting detector CT (PCD-CT) in comparison to energy-integrating CT (EID-CT). Methods Inclusion criteria comprised initial diagnosis of PDAC (reference standard: histopathological analysis) and standardized contrast-enhanced CT imaging either on an EID-CT or a PCD-CT. Patients were excluded due to different histopathological diagnosis or missing tumor delineation on CT. On the PCD-CT, 40–190 keV VMI reconstructions were generated. Image noise, tumor-to-pancreas ratio (TPR) and contrast-to-noise ratio (CNR) were analyzed by ROI-based measurements in arterial and portal venous contrast phase. Two board-certified radiologist evaluated image quality and tumor delineation at both, EID-CT and PCD-CT (40 and 70 keV). Results Thirty-eight patients (mean age 70.4 years ± 10.3 [range 45–91], 27 males; PCD-CT: n=19, EID-CT: n=19) were retrospectively included. On the PCD-CT, tumor conspicuity (reflected by low TPR and high CNR) was significantly improved at low-energy VMI series (≤ 70 keV compared to > 70 keV), both in arterial and in portal venous contrast phase (P < 0.001), reaching the maximum at 40 keV. Comparison between PCD-CT and EID-CT showed significantly higher CNR on the PCD-CT in portal venous contrast phase at < 70 keV (P < 0.016). On the PCD-CT, tumor conspicuity was improved in portal venous contrast phase compared to arterial contrast phase especially at the lower end of the VMI spectrum (≤ 70 keV). Qualitative analysis revealed that tumor delineation is improved in 40 keV reconstructions compared to 70 keV reconstructions on a PCD-CT. Conclusion PCD-CT VMI reconstructions (≤ 70 keV) showed significantly improved conspicuity of PDAC in quantitative and qualitative analysis in both, arterial and portal venous contrast phase, compared to EID-CT, which may be important for early detection of tumor tissue in clinical routine. Tumor delineation was superior in portal venous contrast phase compared to arterial contrast phase. Supplementary Information The online version contains supplementary material available at 10.1007/s00261-023-04042-5.


Introduction
Pancreatic ductal adenocarcinoma (PDAC) remains a leading cause of cancer death; despite advances in diagnosis and treatment in recent years, the 5-year survival rate is low, ranging from 3 to 15% [1][2][3][4][5].Many patients with PDAC present with symptoms such as jaundice or abdominal pain.In most cases, the tumor is detected at a late stage, resulting in unresectable tumor [6].Therefore, early detection of the tumor is a major goal to improve patient survival.
Many imaging modalities are available for the diagnosis of PDAC and have been evaluated in recent years [6].Endoscopic ultrasound is the most sensitive method to detect PDAC [7]; however, it is an invasive method and is often used as an adjunct in clinical routine [6].
Multi-detector computed tomography (MDCT) is the gold standard in imaging of the pancreas [6].It has a high availability and shows a sensitivity of 76-92% and a specificity of 67% for the detection of PDAC [6,[8][9][10].Multiphase CT is the standard imaging technique and has a high sensitivity for detection of PDAC and hepatic metastases [11][12][13].In the pancreatic phase, performed after a delay of about 35-45 s after contrast administration, the tumor can be delineated as a hypoattenuating mass.The portal venous phase (75 s after contrast injection) is important for delineation of metastases, whereas the PDAC itself often shows comparable CT-values to pancreatic tissue [11].
Virtual monoenergetic images (VMI) derived from dualenergy CT (DECT) show promising results for improving abdominal CT imaging, e.g., the better conspicuity of liver metastases at lower keV levels [14,15].The reconstruction of VMI can be performed after image acquisition and is based on the ability of DECT to perform material decomposition.Low keV reconstruction (minimum 40 keV) as well as high keV reconstructions can be carried out.However, low keV reconstructions also show higher image noise which might limit the applicability in clinical routine [16] and could be overcome with noise-optimized VMI (VMI+) [17].At lower keV-levels, the iodine signal is maximized, resulting in a better contrast, but also in higher image noise on DECT [18].Previous studies showed the optimized conspicuity of PDAC in low keV reconstructions using DECT, reaching a maximum at 40 keV [16,[19][20][21][22][23].
Recently, photon-counting detector CT (PCD-CT) has been introduced into clinical routine.These detectors arein contrast to conventional energy-integrating detectors (EID-CT)-capable of directly converting x-ray photons into an electrical signal [24,25].The advantages of this new technology are reduced radiation dose, no electronic noise, improved spatial resolution and intrinsic spectral sensitivity in each scan [24][25][26].These advantages, combined with the known improvement in lesion conspicuity at lower keV levels in DECT, may be a promising tool for the early detection of PDAC.Recent studies showed the improved conspicuity of liver metastases [15] as well as improved subjective image quality for abdominal imaging using low keV reconstructions on a PCD-CT [27].
Therefore, aim of this study was to analyze the potential of PCD-CT in the conspicuity of PDAC at different keV VMI reconstructions and different contrast phases in comparison to EID-CT.

Materials and methods
This retrospective single center study was approved by the local ethics committee and the need to obtain informed consent was waived.The local database was queried for patients with the diagnosis of pancreatic cancer who had undergone contrast-enhanced CT of the (upper) abdomen on a novel Dual-Source Photon Counting CT Scanner (NAEOTOM Alpha, Siemens Healthineers, Erlangen, Germany) as part of routine clinical care between 04/2021 and 07/2021.
The local database was further queried for patients with the diagnosis of pancreatic cancer who had undergone contrast-enhanced CT of the (upper) abdomen on an energyintegrating-detector CT (20-slice MDCT Somatom AS20, Siemens Healthineers, Erlangen, Germany).
Patients' medical charts were reviewed for the following parameters: age, gender, body mass index (kg/m 2 ), tumor pathology and location and CTDI (mGy).Reference standard for the diagnosis of PDAC was histopathological analysis.
Inclusion criteria comprised age ≥ 18 years, contrastenhanced CT of the abdomen either on a PCD-CT or and EID-CT between 04 and 07/2021, histopathological diagnosis of PDAC or in cases with missing histopathological diagnosis clear imaging findings consistent with PDAC.Exclusion criteria comprised other histopathological diagnosis and missing tumor delineation in CT.

Imaging protocol
PCD-CT scans were performed on a dual-source photoncounting detector CT (NAEOTOM Alpha, Siemens Healthineers) as routine clinical acquisitions using a biphasic contrast injection protocol.A contrast bolus of 120 ml (Ultravist 300 mgI/mL, Bayer, Leverkusen, Germany) injected via an antecubital vein was followed by a saline bolus of 30 ml, both using a flow rate of 4.0 ml/s.The scan was bolus-triggered within the ascending aorta (after an attenuation of ≥ 120 HU) with a delay of 20 s for arterial contrast phase and 75 s for portal venous contrast phase.
All patients were scanned craniocaudally in a supine position from the diaphragm or upper thoracic aperture to the pelvis or the symphysis during a single breath-hold.For each scan, we applied the following parameters: acquisition mode with readout of spectral information (QuantumPlus, Siemens Healthineers), 120 kV tube voltage, 0.5 s rot.time, 144 × 0.4 mm collimation.Spectral series were generated using a soft-tissue kernel specifically developed for the spectral postprocessing of PCD-CT datasets (Qr40, QIR 3, Siemens Healthineers) and an enhanced DICOM file format containing spectral information (SPP, spectral postprocessing).Slice thickness was 1.0 mm with an increment of 1.0 mm.
In the comparison group all patients were scanned on an EID-CT (Somatom AS20, Siemens Healthineers) using the following acquisition settings: 120 kV tube voltage, 0.5 s rot.time, 144 × 0.4 mm collimation.On the EID-CT, identical settings for contrast application and delay for image acquisition were used.

Image reconstruction and analysis
Using a dedicated workstation (Syngo.viaVB60A, Siemens Healthineers) we performed post-processing of spectral data.For each patient and for each available contrast phase (arterial phase, portal venous phase), we generated VMI reconstructions (slice thickness of 1 mm, increment of 1mm and identical z-axis orientation) at the following keV levels: 40-80 keV in 5 keV increments and 80-190 keV in 10 keV increments.Post-processing of EID-CT images was also performed using a dedicated workstation (Syngo.viaVB60A).Reconstructions were performed with identical slice thickness of 1 mm and an increment of 1 mm.
Further image analysis was performed using the opensource software Fiji [28], an image processing package based on Image J.For each patient and each available contrast phase, Regions of Interest (ROI) were manually placed in the following dedicated anatomic regions on the 40 keV dataset and automatically copied to all other VMI-datasets; ROI's were placed with the maximum size to correctly describe the lesion/ area: pancreatic tumor tissue (3 ROI's), pancreas (tumor-adjacent pancreatic tissue, 3 ROI's), liver tissue (3 ROI's), aorta, portal vein, inferior vena cava, superior mesenteric artery, superior mesenteric vein, renal cortex (right side), psoas muscle (left side), subcutaneous tissue (right and left side), air (3 ROI's).In total, 21 ROIs were measured at each keV level and at each contrast phase for each patient (Fig. 1).
All ROIs were positioned by a radiologist with 8 years of CT experience.From all ROI's we assessed mean Hounsfield Units (HU) and standard deviation (SD).We calculated median image noise as median of all SD's of all measured ROI's at each keV level/contrast phase and as median of all ROI's measured in subcutaneous tissue.
Tumor-to-pancreas contrast was calculated as ratio between CT-values measured in tumor and in pancreatic parenchyma: This method was performed analogous to the established tumor-to-liver ratio [15,29].Ratios were calculated between all three ROIs per region, resulting in nine tumor-to-pancreas ratios per keV level/contrast phase.Contrast-to-noise ratio (CNR) was calculated as described before:

Qualitative analysis
To assess subjective image quality, two board-certified radiologists reviewed all cases independently; in each case-if available-arterial and portal venous phase, on the PCD-CT 40 and 70 keV reconstructions.Overall image quality as well as tumor delineation were analyzed using a 5-point Likert scale ranging from 1 = poor to 5 = excellent.

Statistical analysis
Analysis of descriptive data and statistical analyses were performed using SPSS 28.0 (IBM Corp. Released 2021.IBM SPSS Statistics for Windows, Version 28.0.Armonk, NY: IBM Corp).Data (CTDI, BMI, noise, CNR, and tumorto-pancreas ratio) are non-normally distributed (shown by Kolmorogov-Smirnov-tests) and therefore presented as median and interquartile range (IQR).Mann-Whitney-U tests were performed to compare different groups.Bonferroni correction was performed for multiple testing.Statistically significant differences were assumed at P values ≤ 0.05.

Patient population
On PCD-CT, 43 patients with a diagnosis of pancreatic cancer were identified from patients with available abdominal CT scan between 04/2021 and 07/2021.Seven patients were excluded due to missing delineation of tumor tissue on the CT scan, six patients were excluded due to previous Whipple procedure, eleven patients were excluded due to different histology after tumor resection (e.g., metastasis, benign tumor), resulting in 19 patients that were included in the study.
On the EID-CT, we identified 21 patients with pancreatic cancer and contrast-enhanced CT scan of the abdomen.Two patients were excluded due to different histology after resection/biopsy, resulting in 19 patients that were included in the study.
In 33/38 patients, pancreatic ductal adenocarcinoma was histologically confirmed either by surgery or biopsy.The remaining five patients had imaging features typical of PDAC (in arterial phase in contrast to pancreatic tissue hypoattenuating mass with lower/comparable CT-values in portal venous phase) [11]; histologic confirmation was not possible due to further treatment at other hospitals (n = 3) and inability to perform biopsy due to poor general health condition or anatomic conditions (n = 2).

Image noise
On the PCD-CT, image noise substantially decreased from a maximum at 40 keV in the arterial contrast phase (26.4 HU [24.8;27.6]) to a median of 13.9 HU [13.4; 15.0] at 90 keV but, did not improve further at higher keV settings (Table 2, Fig. 2).Image noise (measured in subcutaneous tissue) was significantly higher at lower keV-levels (≤ 60 keV) on the PCD-CT compared to the EID-CT, both in arterial and in    2).Similar results were shown for image noise measured in all ROI's (Supplementary Table 1).At lower keV-levels in arterial phase, no significant better tumor conspicuity was observed on the PCD-CT compared to the EID-CT (Table 3).

TPR
Comparison of VMI's at 70 keV on the PCD-CT (which most closely resembles a standard polychromatic 120 kVp acquisition on the EID-CT) with reconstructions on the EID-CT did not show significantly better tumor conspicuity (reflected by TPR) on the PCD-CT, neither in arterial phase nor in portal venous phase (Table 2).
Figure 4 shows VMI reconstructions of a patient with PDAC in the body/tail in portal venous contrast phase.VMI reconstructions in arterial contrast phase are shown in Supplemental Fig. 1.

Contrast phase
No significant differences in TPR were observed between arterial and portal venous contrast phases (Table 4).For all keV VMI reconstructions ≤ 110 keV, CNR was significantly higher in the portal venous contrast phase compared to the arterial contrast phase (e.g., 40 keV: 6.61 [4.39; 9.51] vs. 4.10 [0.94; 4.63], P = 0.016) (Table 4, Fig. 5).The difference in CNR between the arterial and portal venous phases is striking (Fig. 5B), reflecting the high tumor conspicuity at lower keV-levels in portal venous contrast phase compared to arterial contrast phase.

Body Mass Index
In a subgroup analysis, patients were divided according to BMI using a median split (BMI = 22.5 kg/m 2 ).
On the PCD-CT, TPR was significantly better in patients with a BMI < 22.5 kg/m 2 in arterial phase, with however no significant differences in portal venous phase.In contrary, on the EID-CT, TPR was significantly better in patients with a BMI ≥ 22.5 kg/m 2 (P < 0.001) in both phases.CNR was significantly higher in patients with lower BMI on the PCD-CT in arterial phase at higher keV reconstructions, whereas Fig. 4 VMI reconstructions in the portal venous contrast phase in a patient with PDAC in the body/tail of the pancreas.PDAC pancreatic ductal adenocarcinoma CNR was higher for patients with higher BMI on the EID-CT in arterial phase (Supplemental Table 2).

Qualitative image analysis
Comparison between EID-CT and PCD-CT showed improved image quality on a PCD-CT at both contrast phases.However, tumor delineation was not better on a PCD-CT compared to EID-CT at 70 keV reconstructions.Considering only PCD-CT scans, tumor delineation was improved at 40 keV compared to 70 keV in both, arterial and portal venous contrast phase, despite subjective lower image quality (Fig. 6).

Discussion
PCD-CT showed significantly improved tumor conspicuity (reflected by low TPR and high CNR) at lower keV-levels (≤ 70 keV) in arterial and portal venous contrast phases.Compared to EID-CT, tumor delineation on PCD-CT is superior only in the portal venous phase, not in the arterial phase.Tumor tissue showed a slower decrease of CTvalues with increasing keV levels compared to pancreatic tissue in arterial phase, which may also be helpful for the diagnosis.Subjective image analysis showed improved tumor delineation at lower keV levels compared to 70 keV in both, arterial and portal venous phase.
Improved delineation of PDAC at lower keV-levels has previously been shown on dual-energy CT [16,[19][20][21][22]30], but to the best of our knowledge not on a PCD-CT.Best tumor conspicuity as well as best objective and subjective image quality was demonstrated at 40 keV [16].At lower x-ray energies the iodine signal increases resulting in an improved contrast between enhancing tissue (e.g., pancreas) and tissue with reduced contrast enhancement (e.g., PDAC).With increasing availability of PCD-CT, spectral data acquisition is routinely performed on every scan, without increasing radiation dose or dedicated protocols.Previous studies have also highlighted the benefits of low-keV reconstructions in abdominal imaging, demonstrating superior CNR and good subjective image quality [27,31].Improved delineation of hypovascular liver metastases as well as better objective and subjective image quality in oncological imaging have been shown on a novel PCD-CT at lower keV levels (e.g.40 keV) [15,32].
Our study shows higher image noise at lower keV levels.Similar results have been published previously [15,27,33].However, it has been shown for abdominal CT that subjective image quality was higher at lower keV levels despite higher image noise [27].In other studies investigating VMI for other purposes (e.g., vessel delineation), lower keV levels were also preferred by radiologists [32,[34][35][36].Therefore, contrast-to-noise ratio may be considered superior to image noise in the evaluation of contrastenhanced CT scans.
Two phase acquisition (arterial phase and portal venous phase) has been shown to provide the best delineation of PDAC and is therefore routinely performed in patients with suspected pancreatic tumors [37,38].This study shows superior CNR at lower keV levels compared to higher keV levels for both-arterial and venous contrast phases.However, especially at lower keV levels (< 70 keV), CNR was remarkably higher in portal venous   with higher BMI on the EID-CT in both arterial and portal venous phase.On the PCD-CT on the other hand, there were no significant differences between patients with lower and higher BMI in portal venous phase.This may be due to better image quality and lower noise in PCD-CT scans of patients with higher BMI which is most likely caused by the adequate weighting of low-energy photons on PCD-CT.These findings are in line with previous studies on the comparison of abdominal EID-and PCD (low dose) CT where scans of patients with higher BMI showed no significant increase in noise with stable SNR compared with low-BMI scans [15,39].
In this study, two board-certified radiologists independently reviewed the CT scans and rated them on a 5-point Likert scale regarding both, image quality and tumor delineation.Similar results compared to the quantitative analyses were shown.At 70 keV, no better tumor delineation was reported on a PCD-CT compared to the EID-CT.However, at 40 keV radiologists recorded improved tumor delineation compared to 70 keV, both in arterial and portal venous phase.Interestingly, the better subjective tumor delineation was also recorded despite lower subjective image quality at lower keV levels, which might be due to higher image noise.
This study has several limitations.First, the retrospective single-center study design is a major limitation.CT scans were performed during clinical routine; therefore, there are differences in image acquisition protocols between PCD-CT and EID-CT, and not all patients received combined arterial and portal venous phase imaging.Second, the small number of patients is a limitation.However, PDAC is a rare disease.Third, we did not match patients for BMI, age, and gender because of the small number of available patients with an initial diagnosis of PDAC.Therefore, there might be a bias in image noise and CNR due to differences, especially in BMI.However, in the overall patient cohort, BMI and also other demographic parameters were not significantly different.Further, we analyzed different patients that received imaging on different CT scanners and did not perform a head-to-head comparison which might introduce a further bias.Fourth, as the EID-CT does not contain a dual-energy mode, it was not possible to perform direct comparisons between VMI's reconstructed on DECT and PCD-CT.Future studies might consider this to analyze the differences between both VMI's.

Conclusion
Implementation of VMI with low keV levels (e.g.40 keV) for both-arterial and portal venous phase-in clinical routine may improve delineation of pancreatic ductal adenocarcinoma in patients with suspected pancreatic cancer.
Mean HU pancreas − Mean HU tumor SD HU subcutaneous tissue .

Fig. 2
Fig. 2 Median image noise (measured in subcutaneous tissue) on the PCD-CT in arterial (A) and portal venous contrast phase (B).Comparison between image noise on the PCD-CT (70 keV) and the TPR is a measure for delineation of tumor tissue and calculated by the ratio between CT values in tumor tissue and in adjacent pancreatic tissue.Best conspicuity of pancreatic tumor tissue was shown in portal venous phase at the low end of VMI spectrum at 40 keV (TPR = 0.37 [0.20; 0.62]), increasing steadily to a maximum of 0.85 (0.48; 1.05) at 190 keV.TPR was significantly lower for VMI's ≤ 70 keV compared to VMI's > 70 keV, both in portal venous contrast phase (0.44 [0.23; 0.67] vs. 0.70 [0.37; 0.87], P < 0.001) and in arterial contrast phase (0.48 [0.30; 0.83] vs. 0.73 [0.36; 1.05], P < 0.001) (Fig. 3).

Fig. 5
Fig. 5 Tumor-to-pancreas ratio (TPR) (A) and contrast-to-noise ratio (CNR) (B) in comparison between arterial and portal venous contrast phase on the photon-counting detector CT.CT-values of tumor tissue

Fig. 6
Fig. 6 Subjective image analysis.Rating of two board-certified radiologists for image quality and tumor delineation in comparison between EID-CT and PCD-CT in portal venous phase (A, B) and

Table 1
Patient characteristics

Table 3
Tumor-to-pancreas ratio and contrast-to-noise ratio at different keV levels and contrast phases on a PCD-CT and on an EID-CT P < 0.001) and in arterial contrast phase (3.19 [0.70; 3.94] vs. 1

Table 4
Tumor-to-pancreas ratio and contrast-to-noise ratio at different contrast phases on a PCD-CT