Application of deep learning reconstruction of ultra-low-dose abdominal CT in the diagnosis of renal calculi

Zhang, Xiaoxiao; Zhang, Gumuyang; Xu, Lili; Bai, Xin; Zhang, Jiahui; Xu, Min; Yan, Jing; Zhang, Daming; Jin, Zhengyu; Sun, Hansong

doi:10.1186/s13244-022-01300-w

Cited by 8 publications

(7 citation statements)

References 32 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In the study of Zhang et al [ 31 ] which used the same vendor as our study, the mean CT values of the liver parenchyma on non-injected DLR was slightly superior to those of hybrid-IR (57.70 ± 8.36 vs 56.30 ± 8.87, respectively) but the difference was not statistically significant ( P = .389). These results are consistent with our data, as all these studies showed a significantly increased SNR with DLR than with hybrid IR, in addition to better image quality and a lower radiation dose.…”

Section: Discussionmentioning

confidence: 54%

“…Only 2 papers have used the same vendor as the one we used (Canon Medical Systems). [ 31 , 32 ] A possible explanation for this apparent disagreement with the results between all published papers could be the type of vendor used, with its specific DLR algorithm. In the DLR, the quality of the training target determines the performance of the output.…”

Section: Discussionmentioning

confidence: 92%

See 1 more Smart Citation

Impact of a reduced iodine load with deep learning reconstruction on abdominal MDCT

Ludes,

Ohana,

Labani

et al. 2023

Medicine

View full text Add to dashboard Cite

To evaluate the impact of a reduced iodine load using deep learning reconstruction (DLR) on the hepatic parenchyma compared to conventional iterative reconstruction (hybrid IR) and its consequence on the radiation dose and image quality. This retrospective monocentric intraindividual comparison study included 66 patients explored at the portal phase using different multidetector computed tomography parameters: Group A, hybrid IR algorithm (hybrid IR) and a nonionic low-osmolality contrast agent (350 mgI/mL); Group B, DLR algorithm (DLR) and a nonionic iso-osmolality contrast agent (270 mgI/mL). We recorded the attenuation of the liver parenchyma, image quality, and radiation dose parameters. The mean hounsfield units (HU) value of the liver parenchyma was significantly lower in group B, at 105.9 ± 10.9 HU versus 118.5 ± 14.6 HU in group A. However, the 90%IC of mean liver attenuation in the group B (DLR) was between 100.8 HU and 109.3 HU. The signal-to-noise ratio of the liver parenchyma was significantly higher on DLR images, increasing by 56%. However, for both the contrast-to-noise ratio (CNR) and CNR liver/PV no statistical difference was found, even if the CNR liver/PV ratio was slightly higher for group A. The mean dose-length product and computed tomography dose index volume values were significantly lower with DLR, corresponding to a radiation dose reduction of 36% for the DLR. Using a DLR algorithm for abdominal multidetector computed tomography with a low iodine load can provide sufficient enhancement of the liver parenchyma up to 100 HU in addition to the advantages of a higher image quality, a better signal-to-noise ratio and a lower radiation dose.

show abstract

Section: Discussionmentioning

confidence: 54%

Section: Discussionmentioning

confidence: 92%

Impact of a reduced iodine load with deep learning reconstruction on abdominal MDCT

Ludes,

Ohana,

Labani

et al. 2023

Medicine

View full text Add to dashboard Cite

show abstract

“…DLIR has been applied to a wide range of CTA protocols like chest, abdominal and vascular diseases, exhibiting 13.6–37.6% reduction in image noise ( 19 , 31 - 33 ). Our study utilized DLIR-H algorithm to reconstruct one-stop CTA images for the first time and demonstrated its superiority again in preserving image quality under low radiation and contrast dose settings.…”

Section: Discussionmentioning

confidence: 99%

Application of deep learning image reconstruction-high algorithm in one-stop coronary and carotid-cerebrovascular CT angiography with low radiation and contrast doses

Li,

Huang,

et al. 2024

Quant Imaging Med Surg

View full text Add to dashboard Cite

Background For patients with suspected simultaneous coronary and cerebrovascular atherosclerosis, conventional single-site computed tomography angiography (CTA) for both sites can result in nonnegligible radiation and contrast agent dose. The purpose of this study was to validate the feasibility of one-stop coronary and carotid-cerebrovascular CTA (C&CC-CTA) with a “double-low” (low radiation and contrast) dose protocol reconstructed with deep learning image reconstruction with high setting (DLIR-H) algorithm. Methods From February 2018 to January 2019, 60 patients referred to C&CC-CTA simultaneously in West China Hospital were recruited in this prospective cohort study. By random assignment, patients were divided into two groups: double-low dose group (n=30) used 80 kVp and 24 mgI/kg/s contrast dose with images reconstructed using DLIR-H; and routine-dose group (n=30) used 100 kVp and 32 mgI/kg/s contrast dose with images reconstructed using 50% adaptive statistical iterative reconstruction-V (ASIR-V50%). Radiation and contrast doses, subjective image quality score, CT attenuation values, noise, signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) were measured and compared between the groups. Results The DLIR-H group used 30% less contrast dose (35.80±4.85 vs. 51.13±6.91 mL) and 48% less overall radiation dose (1.00±0.09 vs. 1.91±0.42 mSv) than the ASIR-V50% group (both P<0.001). There was no statistically significant difference on subjective quality score between the two groups (C-CTA: 4.38±0.67 vs. 4.17±0.81, P=0.337 and CC-CTA: 4.18±0.87 vs. 4.08±0.79, P=0.604). For coronary CTA, lower background noise (18.93±1.43 vs. 22.86±3.75 HU) was reached in DLIR-H group, and SNR and CNR at all assessed branches were significantly increased compared to ASIR-V50% group (all P<0.05), except SNR of left anterior descending (P>0.05). For carotid-cerebrovascular CTA, DLIR-H group was comparable in background noise (19.25±1.42 vs. 20.23±2.40 HU), SNR and CNR at all assessed branches with ASIR-V50% group (all P>0.05). Conclusions The “double-low” dose one-stop C&CC-CTA with DLIR-H obtained higher image quality compared with the routine-dose protocol with ASIR-V50% while achieving 48% and 30% reduction in radiation and contrast dose, respectively.

show abstract

“…In addition, deep learning has been applied to low-dose or ultra-low-dose CT denoising. Despite the significant dose reduction achieved with ultra-low dose CT, diagnostic performance for urolithiasis remained excellent using two vendor-specific deep learning imaging reconstruction algorithms (TrueFidelity in GE Healthcare and Advanced intelligent Clear-IQ Engine in Cannon Medical System) [ 24 ]. It was consistent with this study in terms of deep learning imaging reconstruction.…”

Section: Discussionmentioning

confidence: 99%

Dose Optimization Using a Deep Learning Tool in Various CT Protocols for Urolithiasis: A Physical Human Phantom Study

Shim,

Choi,

Chang

et al. 2023

Medicina

View full text Add to dashboard Cite

Background and Objectives: We attempted to determine the optimal radiation dose to maintain image quality using a deep learning application in a physical human phantom. Materials and Methods: Three 5 × 5 × 5 mm3 uric acid stones were placed in a physical human phantom in various locations. Three tube voltages (120, 100, and 80 kV) and four current–time products (100, 70, 30, and 15 mAs) were implemented in 12 scans. Each scan was reconstructed with filtered back projection (FBP), statistical iterative reconstruction (IR, iDose), and knowledge-based iterative model reconstruction (IMR). By applying deep learning to each image, we took 12 more scans. Objective image assessments were calculated using the standard deviation of the Hounsfield unit (HU). Subjective image assessments were performed by one radiologist and one urologist. Two radiologists assessed the subjective assessment and found the stone under the absence of information. We used this data to calculate the diagnostic accuracy. Results: Objective image noise was decreased after applying a deep learning tool in all images of FBP, iDose, and IMR. There was no statistical difference between iDose and deep learning-applied FBP images (10.1 ± 11.9, 9.5 ± 18.5 HU, p = 0.583, respectively). At a 100 kV–30 mAs setting, deep learning-applied FBP obtained a similar objective noise in approximately one third of the radiation doses compared to FBP. In radiation doses with settings lower than 100 kV–30 mAs, the subject image assessment (image quality, confidence level, and noise) showed deteriorated scores. Diagnostic accuracy was increased when the deep learning setting was lower than 100 kV–30 mAs, except for at 80 kV–15 mAs. Conclusions: At the setting of 100 kV–30 mAs or higher, deep learning-applied FBP did not differ in image quality compared to IR. At the setting of 100 kV–30 mAs, the radiation dose can decrease by about one third while maintaining objective noise.

show abstract

Application of deep learning reconstruction of ultra-low-dose abdominal CT in the diagnosis of renal calculi

Cited by 8 publications

References 32 publications

Impact of a reduced iodine load with deep learning reconstruction on abdominal MDCT

Impact of a reduced iodine load with deep learning reconstruction on abdominal MDCT

Application of deep learning image reconstruction-high algorithm in one-stop coronary and carotid-cerebrovascular CT angiography with low radiation and contrast doses

Dose Optimization Using a Deep Learning Tool in Various CT Protocols for Urolithiasis: A Physical Human Phantom Study

Contact Info

Product

Resources

About