Reduced-Dose Deep Learning Reconstruction for Abdominal CT of Liver Metastases

Abstract

Background Assessment of liver lesions is constrained as CT radiation doses are lowered; evidence suggests deep learning reconstructions mitigate such effects. Purpose To evaluate liver metastases and image quality between reduced-dose deep learning image reconstruction (DLIR) and standard-dose filtered back projection (FBP) contrast-enhanced abdominal CT. Materials and Methods In this prospective Health Insurance Portability and Accountability Act-compliant study (September 2019 through April 2021), participants with biopsy-proven colorectal cancer and liver metastases at baseline CT underwent standard-dose and reduced-dose portal venous abdominal CT in the same breath hold. Three radiologists detected and characterized lesions at standard-dose FBP and reduced-dose DLIR, reported confidence, and scored image quality. Contrast-to-noise ratios for liver metastases were recorded. Summary statistics were reported, and a generalized linear mixed model was used. Results Fifty-one participants (mean age ± standard deviation, 57 years ± 13; 31 men) were evaluated. The mean volume CT dose index was 65.1% lower with reduced-dose CT (12.2 mGy) than with standard-dose CT (34.9 mGy). A total of 161 lesions (127 metastases, 34 benign lesions) with a mean size of 0.7 cm ± 0.3 were identified. Subjective image quality of reduced-dose DLIR was superior to that of standard-dose FBP (P < .001). The mean contrast-to-noise ratio for liver metastases of reduced-dose DLIR (3.9 ± 1.7) was higher than that of standard-dose FBP (3.5 ± 1.4) (P < .001). Differences in detection were identified only for lesions 0.5 cm or smaller: 63 of 65 lesions detected with standard-dose FBP (96.9%; 95% CI: 89.3, 99.6) and 47 lesions with reduced-dose DLIR (72.3%; 95% CI: 59.8, 82.7). Lesion accuracy with standard-dose FBP and reduced-dose DLIR was 80.1% (95% CI: 73.1, 86.0; 129 of 161 lesions) and 67.1% (95% CI: 59.3, 74.3; 108 of 161 lesions), respectively (P = .01). Lower lesion confidence was reported with a reduced dose (P < .001). Conclusion Deep learning image reconstruction (DLIR) improved CT image quality at 65% radiation dose reduction while preserving detection of liver lesions larger than 0.5 cm. Reduced-dose DLIR demonstrated overall inferior characterization of liver lesions and reader confidence. Clinical trial registration no. NCT03151564 © RSNA, 2022 Online supplemental material is available for this article.

Graphical abstract

Figure 1:

Study flowchart. * = Eight participants were found to have no lesions that fulfilled the study criteria, and two participants were excluded from lesion detection data because they had more than 20 liver lesions. Images from those participants still underwent qualitative and quantitative assessments.

Figure 2:

Box-and-whisker plot shows results of qualitative evaluation of overall image quality rank. A score of 0 was given for the best series, –1 for slightly inferior (no influence on diagnosis), –2 for mildly inferior (possible influence on diagnosis), –3 for moderately inferior (probable influence on diagnosis), and –4 for markedly inferior (impairing diagnosis). Mean image quality rank was significantly different between each reconstruction based on the pairwise comparison (P < .001). Box represents the interquartile range and the whiskers demonstrate the maximum and minimum data range excluding dots that represent outlier data points. The X is the data mean. AV60 = ASIR-V (GE Healthcare) 60%, DLIR = deep learning image reconstruction (with medium strength), FBP = filtered back projection, RD = reduced dose, SD = standard dose.

Figure 3:

Axial contrast-enhanced CT images of the abdomen obtained with standard-dose (SD) filtered back projection (FBP) and reduced-dose (RD) deep learning image reconstruction (DLIR) with medium strength in the same breath hold. A 0.5-cm low-contrast left liver metastasis (arrow with circle) was missed by all three readers at reduced-dose deep DLIR and was detected by all readers at standard-dose FBP. Of note, all three readers qualitatively scored this reduced-dose DLIR scan as a 4 (superior to standard-dose FBP scores of 3 by each reader), even at the aggressive radiation dose reduction of 67% on this scan. Contrast-to-noise ratios for liver metastases in this participant for standard-dose FBP, standard-dose ASIR-V 60% (AV60), standard-dose DLIR, reduced-dose FBP, reduced-dose AV60, and reduced-dose DLIR were 3.6, 4.6, 4.7, 2.1, 3.3, and 3.4, respectively.

Figure 4:

Axial contrast-enhanced CT images show a 0.3-cm liver cyst (arrow) that was detected by all readers on both scans. However, all three reader characterizations were false-positive at standard-dose (SD) filtered back projection (FBP), whereas the regular-dose (RD) deep learning image reconstruction (DLIR) with medium strength scan resulted in two true-negative characterizations and one false-positive characterization. At a radiation dose reduction of 66% on this scan, the cyst appears more conspicuous with DLIR, and each reader qualitatively scored the reduced-dose DLIR scan to be better than or equivalent to the standard-dose FBP scan. Contrast-to-noise ratios for liver metastases in this participant for standard-dose FBP and reduced-dose DLIR were 3.9 and 4.6, respectively.