Single-source dual-energy spectral multidetector CT of pancreatic adenocarcinoma: Optimization of energy level viewing significantly increases lesion contrast

Aim

To evaluate lesion contrast in pancreatic adenocarcinoma patients using spectral multidetector computed tomography (MDCT) analysis.

Materials and methods

The present institutional review board-approved, Health Insurance Portability and Accountability Act of 1996 (HIPAA)-compliant retrospective study evaluated 64 consecutive adults with pancreatic adenocarcinoma examined using a standardized, multiphasic protocol on a single-source, dual-energy MDCT system. Pancreatic phase images (35 s) were acquired in dual-energy mode; unenhanced and portal venous phases used standard MDCT. Lesion contrast was evaluated on an independent workstation using dual-energy analysis software, comparing tumour to non-tumoural pancreas attenuation (HU) differences and tumour diameter at three energy levels: 70 keV; individual subject-optimized viewing energy level (based on the maximum contrast-to-noise ratio, CNR); and 45 keV. The image noise was measured for the same three energies. Differences in lesion contrast, diameter, and noise between the different energy levels were analysed using analysis of variance (ANOVA). Quantitative differences in contrast gain between 70 keV and CNR-optimized viewing energies, and between CNR-optimized and 45 keV were compared using the paired t-test.

Results

Thirty-four women and 30 men (mean age 68 years) had a mean tumour diameter of 3.6 cm. The median optimized energy level was 50 keV (range 40–77). The mean ± SD lesion contrast values (non-tumoural pancreas – tumour attenuation) were: 57 ± 29, 115 ± 70, and 146 ± 74 HU (p = 0.0005); the lengths of the tumours were: 3.6, 3.3, and 3.1 cm, respectively (p = 0.026); and the contrast to noise ratios were: 24 ± 7, 39 ± 12, and 59 ± 17 (p = 0.0005) for 70 keV, the optimized energy level, and 45 keV, respectively. For individuals, the mean ± SD contrast gain from 70 keV to the optimized energy level was 59 ± 45 HU; and the mean ± SD contrast gain from the optimized energy level to 45 keV was 31 ± 25 HU (p = 0.007).

Conclusion

Significantly increased pancreatic lesion contrast was noted at lower viewing energies using spectral MDCT. Individual patient CNR-optimized energy level images have the potential to improve lesion conspicuity.

Introduction

Pancreatic adenocarcinoma remains a leading cause of cancer death in the United States with a continued poor prognosis and high mortality rate.1, 2, 3 Early detection and characterization of pancreatic adenocarcinoma is very important because the 5-year survival rate improves with surgical and/or adjuvant intervention. Multiphasic multidetector computed tomography (MDCT) plays the major role in this work-up.1, 4 Other diagnostic imaging methods, such as endoscopic ultrasound (EUS), magnetic resonance imaging (MRI), occasionally positron-emission tomography (PET), and diagnostic laparoscopy may be used in concert in problematic cases to help characterize and determine whether a cancer is surgically resectable.1, 3, 4, 5

Pancreatic adenocarcinoma is typically hypo-attenuating relative to the adjacent parenchyma and is best depicted on pancreatic-phase MDCT. Complete assessment for vascular invasion and distant metastatic disease also requires hepatic venous phase images.4, 6 Recent retrospective studies suggest that multiphasic MDCT might aid in diagnosis of pancreatic cancers several months before clinical diagnosis.1, 7 However, detection of early disease or elucidation of secondary signs of a cancer continues to pose a diagnostic challenge.⁷ Additionally, there is a subset (4–27%) of pancreatic cancers that is isoattenuating to the adjacent pancreatic parenchyma on state-of-the-art MDCT imaging,⁸ especially when small⁹; these lesions may be notable only by identification of segmental dilation of the pancreatic duct.⁷ Single-source, dual-energy (SSDE) multidetector CT could offer improved ability to detect hypovascular pancreatic adenocarcinomas at lower viewing energy levels during the pancreatic phase of imaging and reduce the number of “underdetected”⁷ early-stage lesions or isoattenuating tumours. However, before application of this spectral approach to the above-described evasive cancers can be performed, it is worthwhile to assess the contribution to lesion conspicuity of typical pancreatic cancers provided by alteration of viewing energies using SSDE MDCT.

The SSDE system utilizes a single x-ray beam source that switches between 80 and 140 kVp at submillisecond intervals during a single helical acquisition. Reprocessing of the data from the x-ray projections allows the creation of simulated mono-energetic image series at 1 keV increments from 40 to 140 keV. Each series can be termed a “viewing energy.” The image set from the simulated 70 keV viewing energy resembles a standard MDCT image obtained with a 120 kVp polychromatic beam. These images sets can viewed in the full range of window/level settings. The ability to instantaneously adjust the monochromatic viewing energies over a range of 40 to 140 keV on the independent workstation of the SSDE system allows spectral imaging of the data, and is different than dual-source, dual-energy MDCT, where weighted averages of the separately acquired 80 and 140 kVp images are combined in image space. With the two commercially available dual-energy techniques available for human use today, the differing acquisition photon energies exploit differences in material composition and, therefore, differences in photon absorption and attenuation. Due to the decreasing photoelectric absorption and increasing Compton scatter that occur with increasing photon energies, the attenuation of iodine is greater at lower energies than at higher energies.10, 11, 12, 13, 14 It follows that the conspicuity of a hypo-attenuating pancreatic mass would be expected to be greater at lower mono-energetic viewing energies when compared to higher energies during pancreatic phase imaging, regardless of single source, dual source, or “low energy” approach to image acquisition. However, image noise is also increased at lower viewing energies (keV), just as it is with lower acquisition energies (kVp). The independent workstation software for the SSDE system can also instantaneously calculate an optimal viewing energy that exploits the difference in iodine concentration of two separate tissues in selected regions of interest based on the contrast-to-noise ratio (CNR) across all available energies from 40–140 keV. This provides an individualized optimal energy for viewing the lesion of interest for each patient. The purpose of the present study was to evaluate lesion contrast in pancreatic adenocarcinoma using spectral MDCT imaging utilizing three different viewing energy levels: 70 keV, individual subject CNR-optimized viewing energy, and 45 keV.