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

Artificial Intelligence in Computed Tomography Image Reconstruction: A Review of Recent Advances

The development of novel image reconstruction algorithms has been pivotal in enhancing image quality and reducing radiation dose in computed tomography (CT) imaging. Traditional techniques like filtered back projection perform well under ideal conditions but fail to generate high-quality images under low-dose, sparse-view, and limited-angle conditions. Iterative reconstruction methods improve upon filtered back projection by incorporating system models and assumptions about the patient, yet they can suffer from patchy image textures. The emergence of artificial intelligence (AI), particularly deep learning, has further advanced CT reconstruction. AI techniques have demonstrated great potential in reducing radiation dose while preserving image quality and noise texture. Moreover, AI has exhibited unprecedented performance in addressing challenging CT reconstruction problems, including low-dose CT, sparse-view CT, limited-angle CT, and interior tomography. This review focuses on the latest advances in AI-based CT reconstruction under these challenging conditions.

Comparing fully automated AI body composition biomarkers at differing virtual monoenergetic levels using dual-energy CT

PURPOSE

To investigate the behavior of artificial intelligence (AI) CT-based body composition biomarkers at different virtual monoenergetic imaging (VMI) levels using dual-energy CT (DECT).

METHODS

This retrospective study included 88 contrast-enhanced abdominopelvic CTs acquired with rapid-kVp switching DECT. Images were reconstructed into five VMI levels (40, 55, 70, 85, 100 keV). Fully automated algorithms for quantifying CT number (HU) in abdominal fat (subcutaneous and visceral), skeletal muscle, bone, calcium (abdominal Agatston score), and organ size (area or volume) were applied. Biomarker median difference relative to 70 keV and interquartile range were reported by energy level to characterize variation. Linear regression was performed to calibrate non-70 keV data and to estimate their equivalent 70 keV biomarker attenuation values.

RESULTS

Relative to 70 keV, absolute median differences in attenuation-based biomarkers (excluding Agatston score) ranged 39-358, 12-102, 5-48, 9-75 HU for 40, 55, 85, 100 keV, respectively. For area-based biomarkers, differences ranged 6-15, 3-4, 2-7, 0-5 cm2 for 40, 55, 85, 100 keV. For volume-based biomarkers, differences ranged 12-34, 8-68, 12-52, 1-57 cm3 for 40, 55, 85, 100 keV. Agatston score behavior was more spurious with median differences ranging 70-204 HU. In general, VMI < 70 keV showed more variation in median biomarker measurement than VMI > 70 keV.

CONCLUSION

This study characterized the behavior of a fully automated AI CT biomarker toolkit across varying VMI levels obtained with DECT. The data showed relatively little biomarker value change when measured at or greater than 70 keV. Lower VMI datasets should be avoided due to larger deviations in measured value as compared to 70 keV, a level considered equivalent to conventional 120 kVp exams.

Adjustments of iodinated contrast media using lean body weight for abdominopelvic computed tomography: A systematic review and meta-analysis

PURPOSE

This systematic review aimed to compare the effect of contrast media (CM) dose adjustment based on lean body weight (LBW) method versus other calculation protocols for abdominopelvic CT examinations.

METHOD

Studies published from 2002 onwards were systematically searched in June 2024 across Medline, Embase, CINAHL, Cochrane CENTRAL, Web of Science, Google Scholar and four other grey literature sources, with no language limit. Randomised controlled trials (RCT) and quasi-RCT of abdominopelvic or abdominal CT examinations in adults with contrast media injection for oncological and acute diseases were included. The comparators were other contrast dose calculation methods such as total body weight (TBW), fixed volume (FV), body surface area (BSA), and blood volume. The main outcomes considered were liver and aortic enhancement. Titles, abstracts and full texts were independently screened by two reviewers.

RESULTS

Eight studies were included from a total of 2029 articles identified. Liver parenchyma and aorta contrast enhancement did not significantly differ between LBW and TBW protocols (p = 0.07, p = 0.06, respectively). However, the meta-analysis revealed significantly lower contrast volume injected with LBW protocol when compared to TBW protocol (p = 0.003). No statistical differences were found for contrast enhancement and contrast volume between LBW and the other strategies.

CONCLUSION

Calculation of the CM dosage based on LBW allows a reduction in the injected volume for abdominopelvic CT examination, ensuring the same image quality in terms of contrast enhancement.

Prediction of proton stopping power ratios using dual-energy CT basis material decomposition

BACKGROUND

Proton radiotherapy treatment plans are currently restricted by the range uncertainties originating from the stopping power ratio (SPR) prediction based on single-energy computed tomography (SECT). Various studies have shown that multi-energy CT (MECT) can reduce the range uncertainties due to medical implant materials and age-related variations in tissue composition. None of these has directly applied the basis material density (MD) images produced by projection-based MECT systems for SPR prediction.

PURPOSE

To present and evaluate a novel proton SPR prediction method based on MD images from dual-energy CT (DECT), which could reduce the range uncertainties currently associated with proton radiotherapy.

METHODS

A theoretical basis material decomposition into water and iodine material densities was performed for various pediatric and adult human reference tissues, as well as other non-tissue materials, by minimizing the root-mean-square relative attenuation error in the energy interval from 40 to 140 keV. A model (here called MD-SPR) mapping predicted MDs to theoretically calculated reference SPRs was created with locally weighted scatterplot smoothing (LOWESS) data-fitting. The goodness of fit of the MD-SPR model was evaluated for the included reference tissues. MD images of two electron density phantoms, combined to form a head- and an abdomen-sized phantom setup, were acquired with a clinical projection-based fast-kV switching DECT scanner. The MD images were compared to the theoretically predicted MDs of the tissue surrogates and other non-tissue materials in the phantoms, as well as used for input to the MD-SPR model for generation of SPR images. The SPR images were subsequently compared to theoretical reference SPRs of the materials in the phantoms, as well as to SPR images from a commercial algorithm (DirectSPR, Siemens Healthineers, Forchheim, Germany) using image-based consecutive scan DECT for the same phantom setups.

RESULTS

The predicted SPRs of the tissue surrogates were similar for MD-SPR and DirectSPR, where the adipose and bone tissue surrogates were within 1% difference to the reference SPRs, while other non-adipose soft tissue surrogates (breast, brain, liver, muscle) were all underestimated by between -0.7% and -1.8%. The SPRs of the non-tissue materials (polymethyl methacrylate (PMMA), polyether ether ketone (PEEK), graphite and Teflon) were within 2.8% for MD-SPR images, compared to 6.8% for DirectSPR.

CONCLUSIONS

The MD-SPR model performed similar compared to other published methods for the human reference tissues. The SPR prediction for tissue surrogates was similar to DirectSPR and showed potential to improve SPR prediction for non-tissue materials.