Deep learning image reconstruction (DLIR) may facilitate up to a 45 percent reduction in radiation dosing with the use of dual-energy computed tomography (DECT) for detection of liver metastases in comparison to single-energy CT (SECT) with standard dosing, according to a new study.
For the prospective study, recently published in the European Journal of Radiology, researchers compared SECT with standard radiation dosing (120-kVp images with adaptive statistical iterative reconstruction-Veo at 40 percent (AR40) versus DECT with reduced radiation dosing (40- and 70-keV virtual monoenergetic images (VMIs)) with high-intensity DLIR. The cohort was comprised of 80 patients with a mean age of 59 and mean body mass index (BMI) of 22.56 kg/m2, according to the study. The researchers noted that all study participants had known or suspected liver metastases.
The study authors found that in comparison to AR40 120-kVp with SECT, the 40-keV and 70-keV dosing with DECT and high-intensity DLIR (DH) offered significantly higher contrast-to-noise ratios (CNRs) in the liver (9.66 and 5.40 respectively vs. 3.85) and portal vein (25.89 and 12.66 respectively vs. 7.33). The 40-keV and 70-keV dosing groups also demonstrated higher signal-to-noise ratios (SNRs) versus AR40 120-kVp in the liver (10.53 and 9.81 vs. 7.45) and the portal vein (14.36 and 12.05 vs. 9.27), according to the researchers.
The 40-keV dosing provided more than double the liver to lesion contrast to noise ratio (LLR) of the AR40 120-kVp dosing (8.86 vs. 4.07) and the 70-keV dosing offered a 6.42 LLR. However, the study authors pointed out that the mean volume CT dose index (CTDIvol) was 10.66 mGy for the SECT exams with standard dosing in comparison to 5.63 mGy for the DECT cohort.
“Our study demonstrated that, at a 45 % radiation dose reduction, the DH 40- and 70-keV VMIs achieved improved CNR and SNR, comparable or improved image quality, and superior or similar conspicuity of liver metastases, and maintained equivalent lesion detection rates compared to standard-dose AR40 120-kVp images,” wrote lead study author Yuncheng Li, M.D., who is affiliated with the Department of Radiology at the First Affiliated Hospital of Anhui Medical University in Anhui, China, and colleagues.
The study authors noted a 97.06 percent lesion detection rate with AR40 120-kVp imaging in contrast to 97.84 percent for DH 40 keV VMIs and 95.88 percent for DH 70-keV VMIs.
Three Key Takeaways
- Radiation dose reduction with preserved diagnostic accuracy. Dual-energy CT (DECT) with deep learning image reconstruction (DLIR) enabled a 45 percent reduction in radiation dose while maintaining comparable or improved image quality and lesion detection rates versus standard-dose single-energy CT (SECT).
- Improved image quality and lesion conspicuity. DECT with DLIR at 40- and 70-keV virtual monoenergetic images achieved higher contrast-to-noise and signal-to-noise ratios with superior liver-to-lesion conspicuity compared to standard SECT.
- Reliable lesion characterization with spectral parameters. Spectral DECT metrics such as spectral slope and effective atomic number (Zeff) mapping provided high accuracy in differentiating cysts from metastases, supporting robust quantitative assessment even at reduced doses.
Additionally, the researchers demonstrated that spectral DECT parameters had robust accuracy in differentiating between cysts and metastatic lesions. Spectral slope provided a 97.7 percent AUC and 94 percent sensitivity and effective atomic number mapping (Zeff) offered a 98 percent AUC and 89 percent sensitivity, according to the study authors.
“By preserving image quality, DLIR ensures these quantitative spectral measurements stay reliable even at low dose, allowing confident lesion characterization. In fact, we speculate that low-dose DECT with DLIR may outperform some standard-dose protocols in overall diagnostic performance, since it provides stable, high-contrast images with accurate quantitative data,” added Li and colleagues.
(Editor’s note: For related content, see “What New Interventional Radiology Research Reveals About Treatment for Breast Cancer Liver Metastases,” “Abbreviated MRI for Hepatocellular Carcinoma: What a New Meta-Analysis Reveals” and “New CT and MRI Research Shows Link Between LR-M Lesions and Rapid Progression of Early-Stage HCC.”)
In regard to study limitations, the authors acknowledged the relatively small cohort size, the use of one vendor’s DECT system, the lack of histopathological confirmation for all lesions and the emphasis on assessing images only in the portal venous phase.
