A novel fast kilovoltage switching dual-energy CT with deep learning: Accuracy of CT number on virtual monochromatic imaging and iodine quantification

Systematic Review on Learning-based Spectral CT

Spectral computed tomography (CT) has recently emerged as an advanced version of medical CT and significantly improves conventional (single-energy) CT. Spectral CT has two main forms: dual-energy computed tomography (DECT) and photon-counting computed tomography (PCCT), which offer image improvement, material decomposition, and feature quantification relative to conventional CT. However, the inherent challenges of spectral CT, evidenced by data and image artifacts, remain a bottleneck for clinical applications. To address these problems, machine learning techniques have been widely applied to spectral CT. In this review, we present the state-of-the-art data-driven techniques for spectral CT.

Performance evaluation of deep learning image reconstruction algorithm for dual-energy spectral CT imaging: A phantom study

OBJECTIVES

To evaluate the performance of deep learning image reconstruction (DLIR) algorithm in dual-energy spectral CT (DEsCT) as a function of radiation dose and image energy level, in comparison with filtered-back-projection (FBP) and adaptive statistical iterative reconstruction-V (ASIR-V) algorithms.

METHODS

An ACR464 phantom was scanned with DEsCT at four dose levels (3.5 mGy, 5 mGy, 7.5 mGy, and 10 mGy). Virtual monochromatic images were reconstructed at five energy levels (40 keV, 50 keV, 68 keV, 74 keV, and 140 keV) using FBP, 50% and 100% ASIR-V, DLIR at low (DLIR-L), medium (DLIR-M), and high (DLIR-H) settings. The noise power spectrum (NPS), task-based transfer function (TTF) and detectability index (d') were computed and compared among reconstructions.

RESULTS

NPS area and noise increased as keV decreased, with DLIR having slower increase than FBP and ASIR-V, and DLIR-H having the lowest values. DLIR had the best 40 keV/140 keV noise ratio at various energy levels, DLIR showed higher TTF (50%) than ASIR-V for all materials, especially for the soft tissue-like polystyrene insert, and DLIR-M and DLIR-H provided higher d' than DLIR-L, ASIR-V and FBP in all dose and energy levels. As keV increases, d' increased for acrylic insert, and d' of the 50 keV DLIR-M and DLIR-H images at 3.5 mGy (7.39 and 8.79, respectively) were higher than that (7.20) of the 50 keV ASIR-V50% images at 10 mGy.

CONCLUSIONS

DLIR provides better noise containment for low keV images in DEsCT and higher TTF(50%) for the polystyrene insert over ASIR-V. DLIR-H has the lowest image noise and highest detectability in all dose and energy levels. DEsCT 50 keV images with DLIR-M and DLIR-H show potential for 65% dose reduction over ASIR-V50% withhigher d'.

Dual-Energy Computed Tomography: Technological Considerations

Compared to conventional single-energy CT (SECT), dual-energy CT (DECT) provides additional information to better characterize imaged tissues. Approaches to DECT acquisition vary by vendor and include source-based and detector-based systems, each with its own advantages and disadvantages. Despite the different approaches to DECT acquisition, the most utilized DECT images include routine SECT equivalent, virtual monoenergetic, material density (eg, iodine map), and virtual non-contrast images. These images are generated either through reconstructions in the projection or image domains. Designing and implementing an optimal DECT workflow into routine clinical practice depends on radiologist and technologist input with special considerations including appropriate patient and protocol selection and workflow automation. In addition to better tissue characterization, DECT provides numerous advantages over SECT such as the characterization of incidental findings and dose reduction in radiation and iodinated contrast.

Effects of scan parameters on the accuracies of iodine quantification and hounsfield unit values in dual layer dual-energy head and neck computed tomography: A phantom study conducted in a hospital in Japan

INTRODUCTION

No study has investigated scan parameters in head and neck dual layer dual-energy computed tomography (DL-DECT). This study aimed to select the appropriate scan parameters in head and neck imaging by evaluating the scan parameter effects on the accuracies of CT numbers and conduct iodine quantification in DL-DECT.

METHODS

A multi-energy phantom was scanned using a dual layer CT (DLCT) scanner. Reference materials of iodine, blood, calcium, and adipose were used. A helical scan was performed by using reference and several protocols. Iodine density and virtual monochromatic images (VMIs) at the energy of 50, 70, and 100 keV were reconstructed. The iodine concentrations and CT numbers in each protocol were measured. Moreover, the absolute percentage errors (APEs) of iodine quantifications and CT numbers (reference vs. each protocol) were compared. Equivalence was observed when APEs between reference and each protocol was within 5%. Statistical analysis was performed using appropriate software.

RESULTS

The APEs between the high-tube-voltage and reference protocol were 23.7, 14.0, 8.8, and 8.1% for iodine reference materials with concentrations equal to 2, 5, 10, and 15 mg/ml, respectively. At 50 keV, APEs between the high-tube-voltage and reference protocols were greater than 5% except for calcium and adipose. At 100 keV, APEs between the high-tube-voltage and reference protocols were greater than 5% except for blood and calcium.

CONCLUSIONS

The high-tube-voltage protocol improved the accuracies of the measurement for iodine quantification and CT numbers. Additionally, the scanning parameters except for tube voltage had no effect on accuracies of iodine quantitation and CT numbers in the DLCT scanner.

IMPLICATIONS FOR PRACTICE

The use of the high-tube-voltage protocol will be recommended for more accurate material decomposition in head and neck DL-DECT.

Impact of the automatic tube current modulation (ATCM) system on virtual monoenergetic image quality for dual-source CT: A phantom study

PURPOSE

To assess the impact of the automatic tube current modulation (ATCM) on virtual monoenergetic images (VMIs) quality in dual-source CT(DSCT).

MATERIALS AND METHODS

Acquisitions were performed on DSCT using the Mercury phantom. The acquisition parameters for an abdomen-pelvic examination with single-energy CT(SECT) and dual-energy CT(DECT) imaging were used. Acquisitions were performed for each imaging mode using fixed mAs and ATCM. The mAs value was set to obtain a volume CT dose index of 11 mGy in fixed mAs acquisitions. This value was used as the reference mAs in ATCM acquisitions. The noise power spectrum and task-based transfer function at 40,50,60 and 70 keV levels were computed on VMIs and SECT images. The detectability index (d') was calculated for a lesion with an iodine concentration of 10 mg/mL.

RESULTS

The noise magnitude on VMIs was higher with the ATCM system than with fixed mAs for all energy levels and section diameters of 21,26 and 31 cm. The noise texture and spatial resolution were similar between the fixed mAs and ATCM acquisitions for both imaging modes. The d' values were lower for all energy levels with ATCM than with fixed mAs acquisitions for 21 and 26 cm diameters by -39.82 ± 9.32%, similar at 31 cm diameter -4.13 ± 0.24% and higher at 36 cm diameter 10.40 ± 6.69%. It was higher on VMIs at all energy levels compared to SECT images.

CONCLUSIONS

The ATCM system could be used with DECT imaging to optimize patient exposure without changing the noise texture and spatial resolution of VMIs compared to fixed mAs and SECT.

Effect of iodine concentration and body size on iodine subtraction in virtual non-contrast imaging: A phantom study

INTRODUCTION

Dual-energy computed tomography (DECT) can generate virtual non-contrast (VNC) images. Herein, we sought to improve the accuracy of VNC images by identifying the optimal slope of contrast media (SCM) for VNC-image generation based on the iodine concentration and subject's body size.

METHODS

We used DECT to scan a multi-energy phantom including four iodine concentration rods (15, 10, 5, and 2 mg/mL), and 240 VNC images (eight SCM ranging from 0.49 to 0.56 × three body sizes × ten scans) that were generated by three-material decomposition. The CT number of each iodine and solid water rod part was measured in each VNC image. The difference in the CT number between the iodine and the solid water rod part was calculated and compared using paired t-test or repeated measures ANOVA.

RESULTS

The SCM that achieved an absolute value of the difference in CT number of <5.0 Hounsfield units (HU) for all body sizes simultaneously was greater at lower iodine concentration (SCM of 0.5, 0.51, and 0.53 at 10, 5, and 2 mg/mL iodine, respectively). At an iodine concentration of 15 mg/mL, no SCM achieved an absolute difference of <5.0 HU in CT number for all body sizes simultaneously. At all iodine concentrations, the SCM achieving the minimal difference in the CT number increased with the increase in body size.

CONCLUSION

By adjusting the SCM according to the iodine concentration and body size, it is possible to generate VNC images with an accuracy of <5.0 HU.

IMPLICATIONS FOR PRACTICE

Improving the accuracy of VNC images minimizing incomplete iodine subtraction would make it possible to replace true non-contrast (TNC) images with VNC images and reduce the radiation dose.

A Characterization of Deep Learning Reconstruction Applied to Dual-Energy Computed Tomography Monochromatic and Material Basis Images

OBJECTIVE

Advancements in computed tomography (CT) reconstruction have enabled image quality improvements and dose reductions. Previous advancements have included iterative and model-based reconstruction. The latest image reconstruction advancement uses deep learning, which has been evaluated for polychromatic imaging only. This article characterizes a commercially available deep learning imaging reconstruction applied to dual-energy CT.

METHODS

Monochromatic, iodine basis, and water basis images were reconstructed with filtered back projection (FBP), iterative (ASiR-V), and deep learning (DLIR) methods in a phantom experiment. Slice thickness, contrast-to-noise ratio, modulation transfer function, and noise power spectrum metrics were used to characterize ASiR-V and DLIR relative to FBP over a range of dose levels, phantom sizes, and iodine concentrations.

RESULTS

Slice thicknesses for ASiR-V and DLIR demonstrated no statistically significant difference relative to FBP for all measurement conditions. Contrast-to-noise ratio performance for DLIR-high and ASiR-V 40% at 2 mg I/mL on 40-keV images were 162% and 30% higher than FBP, respectively. Task-based modulation transfer function measurements demonstrated no clinically significant change between FBP and ASiR-V and DLIR on monochromatic or iodine basis images.

CONCLUSIONS

Deep learning image reconstruction enabled better image quality at lower monochromatic energies and on iodine basis images where image contrast is maximized relative to polychromatic or high-energy monochromatic images. Deep learning image reconstruction did not demonstrate thicker slices, decreased spatial resolution, or poor noise texture (ie, "plastic") relative to FBP.

Iodine maps derived from sparse-view kV-switching dual-energy CT equipped with a deep learning reconstruction for diagnosis of hepatocellular carcinoma

Deep learning-based spectral CT imaging (DL-SCTI) is a novel type of fast kilovolt-switching dual-energy CT equipped with a cascaded deep-learning reconstruction which completes the views missing in the sinogram space and improves the image quality in the image space because it uses deep convolutional neural networks trained on fully sampled dual-energy data acquired via dual kV rotations. We investigated the clinical utility of iodine maps generated from DL-SCTI scans for assessing hepatocellular carcinoma (HCC). In the clinical study, dynamic DL-SCTI scans (tube voltage 135 and 80 kV) were acquired in 52 patients with hypervascular HCCs whose vascularity was confirmed by CT during hepatic arteriography. Virtual monochromatic 70 keV images served as the reference images. Iodine maps were reconstructed using three-material decomposition (fat, healthy liver tissue, iodine). A radiologist calculated the contrast-to-noise ratio (CNR) during the hepatic arterial phase (CNR_a) and the equilibrium phase (CNR_e). In the phantom study, DL-SCTI scans (tube voltage 135 and 80 kV) were acquired to assess the accuracy of iodine maps; the iodine concentration was known. The CNR_a was significantly higher on the iodine maps than on 70 keV images (p < 0.01). The CNR_e was significantly higher on 70 keV images than on iodine maps (p < 0.01). The estimated iodine concentration derived from DL-SCTI scans in the phantom study was highly correlated with the known iodine concentration. It was underestimated in small-diameter modules and in large-diameter modules with an iodine concentration of less than 2.0 mgI/ml. Iodine maps generated from DL-SCTI scans can improve the CNR for HCCs during hepatic arterial phase but not during equilibrium phase in comparison with virtual monochromatic 70 keV images. Also, when the lesion is small or the iodine concentration is low, iodine quantification may result in underestimation.

Characteristics of the deep learning-based virtual monochromatic image with fast kilovolt-switching CT: a phantom study

PURPOSE

We assessed the physical properties of virtual monochromatic images (VMIs) obtained with different energy levels in various contrast settings and radiation doses using deep learning-based spectral computed tomography (DL-Spectral CT) and compared the results with those from single-energy CT (SECT) imaging.

MATERIALS AND METHODS

A Catphan^® 600 phantom was scanned by DL-Spectral CT at various radiation doses. We reconstructed the VMIs obtained at 50, 70, and 100 keV. SECT (120 kVp) images were acquired at the same radiation doses. The standard deviations of the CT number and noise power spectrum (NPS) were calculated for noise characterization. We evaluated the spatial resolution by determining the 10% task-based transfer function (TTF) level, and we assessed the task-based detectability index (d').

RESULTS

Regardless of the radiation dose, the noise was the lowest at 70 keV VMI. The NPS showed that the noise amplitude at all spatial frequencies was the lowest among other VMI and 120 kVp images. The spatial resolution was higher for 70 keV VMI compared to the other VMIs, except for high-contrast objects. The d' of 70 keV VMI was the highest among the VMI and 120 kVp images at all radiation doses and contrast settings. The d' of the 70 keV VMIs at the minimum dose was higher than that at the maximum dose in any other image.

CONCLUSION

The physical properties of the DL-Spectral CT VMIs varied with the energy level. The 70 keV VMI had the highest detectability by far among the VMI and 120-kVp images. DL-Spectral CT may be useful to reduce radiation doses.

Deep learning image reconstruction to improve accuracy of iodine quantification and image quality in dual-energy CT of the abdomen: a phantom and clinical study

OBJECTIVES

To investigate the effect of deep learning image reconstruction (DLIR) on the accuracy of iodine quantification and image quality of dual-energy CT (DECT) compared to that of other reconstruction algorithms in a phantom experiment and an abdominal clinical study.

METHODS

An elliptical phantom with five different iodine concentrations (1-12 mgI/mL) was imaged five times with fast-kilovoltage-switching DECT for three target volume CT dose indexes. All images were reconstructed using filtered back-projection, iterative reconstruction (two levels), and DLIR algorithms. Measured and nominal iodine concentrations were compared among the algorithms. Contrast-enhanced CT of the abdomen with the same scanner was acquired in clinical patients. In arterial and portal venous phase images, iodine concentration, image noise, and coefficients of variation for four locations were retrospectively compared among the algorithms. One-way repeated-measures analyses of variance were used to evaluate differences in the iodine concentrations, standard deviations, coefficients of variation, and percentages of error among the algorithms.

RESULTS

In the phantom study, the measured iodine concentrations were equivalent among the algorithms: within ± 8% of the nominal values, with root-mean-square deviations of 0.08-0.36 mgI/mL, regardless of radiation dose. In the clinical study (50 patients; 35 men; mean age, 68 ± 11 years), iodine concentrations were equivalent among the algorithms for each location (all p > .99). Image noise and coefficients of variation were lower with DLIR than with the other algorithms (all p < .01).

CONCLUSIONS

The DLIR algorithm reduced image noise and variability of iodine concentration values compared with other reconstruction algorithms in the fast-kilovoltage-switching dual-energy CT.

KEY POINTS

• In the phantom study, standard deviations and coefficients of variation in iodine quantification were lower on images with the deep learning image reconstruction algorithm than on those with other algorithms. • In the clinical study, iodine concentrations of measurement location in the upper abdomen were consistent across four reconstruction algorithms, while image noise and variability of iodine concentrations were lower on images with the deep learning image reconstruction algorithm.

Comparison of image quality of two versions of deep-learning image reconstruction algorithm on a rapid kV-switching CT: a phantom study

BACKGROUND

To assess the impact of the new version of a deep learning (DL) spectral reconstruction on image quality of virtual monoenergetic images (VMIs) for contrast-enhanced abdominal computed tomography in the rapid kV-switching platform.

METHODS

Two phantoms were scanned with a rapid kV-switching CT using abdomen-pelvic CT examination parameters at dose of 12.6 mGy. Images were reconstructed using two versions of DL spectral reconstruction algorithms (DLSR V1 and V2) for three reconstruction levels. The noise power spectrum (NSP) and task-based transfer function at 50% (TTF₅₀) were computed at 40/50/60/70 keV. A detectability index (d') was calculated for enhanced lesions at low iodine concentrations: 2, 1, and 0.5 mg/mL.

RESULTS

The noise magnitude was significantly lower with DLSR V2 compared to DLSR V1 for energy levels between 40 and 60 keV by -36.5% ± 1.4% (mean ± standard deviation) for the standard level. The average NPS frequencies increased significantly with DLSR V2 by 23.7% ± 4.2% for the standard level. The highest difference in TTF₅₀ was observed at the mild level with a significant increase of 61.7% ± 11.8% over 40-60 keV energy with DLSR V2. The d' values were significantly higher for DLSR V2 versus DLSR V1.

CONCLUSIONS

The DLSR V2 improves image quality and detectability of low iodine concentrations in VMIs compared to DLSR V1. This suggests a great potential of DLSR V2 to reduce iodined contrast doses.

Comparison of virtual monoenergetic imaging between a rapid kilovoltage switching dual-energy computed tomography with deep-learning and four dual-energy CTs with iterative reconstruction

Background

To assess the spectral performance of rapid kV switching dual-energy CT (KVSCT-Canon) equipped with a Deep-Learning spectral reconstruction algorithm on virtual-monoenergetic images at low-energy levels and to compare its performances with four other dual-energy CT (DECT) platforms equipped with iterative reconstruction algorithms.

Methods

Two CT phantoms were scanned on five DECT platforms: KVSCT-Canon, fast kV-switching CT (KVSCT-GE), split filter CT, dual-source CT (DSCT), and dual-layer CT (DLCT). The classical parameters of abdomen-pelvic examinations were used for all phantom acquisitions, and a CTDI_vol close to 10 mGy. For KVSCT-Canon, virtual-monoenergetic images were reconstructed with a clinical slice thickness of 0.5 and 1.5 mm to be close to other platforms. Noise power spectrum (NPS) and task-based transfer function (TTF) were evaluated from 40 to 80 keV of virtual-monoenergetic images. A detectability index (d') was computed to model the detection task of two contrast-enhanced lesions as function of keV.

Results

For KVSCT-Canon, the noise magnitude and average NPS spatial frequency (f_av) decreased from 40 to 70 keV and increased thereafter. Similar noise magnitude outcomes were found for KVSCT-GE but the opposite for f_av. For the other DECT platforms, the noise magnitude decreased as the keV increased. For split filter CT, DSCT and DLCT, the f_av values increased from 40 to 80 keV. For all DECT platforms, TTF at 50% (f₅₀) decreased as the keV increased, decreasing spatial resolution. For KVSCT-Canon, d' values peaked at 60 and 70 keV for both simulated lesions and from 50 to 70 keV for KVSCT-GE. d' decreased between 40 and 70 keV for DSCT, DLCT and split filter CT. For KVSCT-Canon, the increase in slice thickness decreases noise magnitude, f_av and f₅₀ and increases d' values. The highest d' values were found for DLCT at 40 and 50 keV and for KVSCT-Canon at 1.5 mm for other keV.

Conclusions

For KVSCT-Canon, the detectability of contrast-enhanced lesions was highest at 60 keV. The highest d' values were found for DLCT at 40 and 50 keV and for KVSCT-Canon at 1.5 mm for other keV.