Effect of a new deep learning image reconstruction algorithm for abdominal computed tomography imaging on image quality and dose reduction compared with two iterative reconstruction algorithms: a phantom study

Purpose High radiosensitivity of children undergoing repetitive computed tomography examinations necessitates the use of iterative reconstruction algorithms in order to achieve a significant radiation dose reduction. The goal of this study is to compare the iDose iterative reconstruction algorithm with filtered backprojection in terms of radiation exposure and image quality in 33 chest high-resolution computed tomography examinations performed in young children with chronic bronchitis. Methods Fourteen patients were scanned using the filtered backprojection protocol while 19 patients using the iDose protocol and reduced milliampere-seconds, both on a 64-detector row computed tomography scanner. The iDose group images were reconstructed with different iDose levels (2, 4, and 6). Radiation exposure quantities were estimated, while subjective and objective image qualities were evaluated. Unpaired t tests were used for data statistical analysis. Results The iDose application allowed significant effective dose reduction (about 80%). Subjective image quality evaluation showed satisfactory results even with iDose level 2, whereas it approached excellent image with iDose level 6. Subjective image noise was comparable between the 2 groups with the use of iDose level 4, while objective noise was comparable between filtered backprojection and iterative reconstruction level 6 images. Conclusions The iDose algorithm use in pediatric chest high-resolution computed tomography reduces radiation exposure without compromising image quality. Further evaluation with iterative reconstruction algorithms is needed in order to establish high-resolution computed tomography as the gold standard low-dose method for children suffering from chronic lung diseases.

Download Full-text

Assessing image quality and dose reduction of a new x-ray computed tomography iterative reconstruction algorithm using model observers

Medical Physics ◽

10.1118/1.4881143 ◽

2014 ◽

Vol 41 (7) ◽

pp. 071910 ◽

Cited By ~ 22

Author(s):

Hsin-Wu Tseng ◽

Jiahua Fan ◽

Matthew A. Kupinski ◽

Paavana Sainath ◽

Jiang Hsieh

Keyword(s):

Computed Tomography ◽

Image Quality ◽

Dose Reduction ◽

Iterative Reconstruction ◽

Reconstruction Algorithm ◽

X Ray ◽

Iterative Reconstruction Algorithm ◽

X Ray Computed ◽

Model Observers

Download Full-text

CT pulmonary angiography: dose reduction via a next generation iterative reconstruction algorithm

Acta Radiologica ◽

10.1177/0284185118784976 ◽

2018 ◽

Vol 60 (4) ◽

pp. 478-487 ◽

Cited By ~ 7

Author(s):

Andreas Sauter ◽

Thomas Koehler ◽

Bernhard Brendel ◽

Juliane Aichele ◽

Jan Neumann ◽

...

Keyword(s):

Image Quality ◽

Dose Reduction ◽

Iterative Reconstruction ◽

Low Dose ◽

Reconstruction Algorithm ◽

Pulmonary Angiography ◽

Reconstruction Algorithms ◽

Iterative Reconstruction Algorithm ◽

Dose Levels ◽

Rule Out

Background Computed tomography pulmonary angiography (CTPA) is the standard imaging modality for detection or rule out of pulmonary embolism (PE); however, radiation exposure is a serious concern. With iterative reconstruction algorithms a distinct dose reduction could be achievable. Purpose To evaluate a next generation iterative reconstruction algorithm for detection or rule-out of PE in simulated low-dose CTPA. Material and Methods Low-dose CT datasets with 50%, 25%, and 12.5% of the original tube current were simulated based on CTPA examinations of 92 patients with suspected PE. All datasets were reconstructed with two reconstruction algorithms: standard filtered back-projection (FBP) and iterative model reconstruction (IMR). In total, 736 CTPA datasets were evaluated by three blinded radiologists regarding image quality, diagnostic confidence, and detectability of PE. Furthermore, contrast-to-noise ratio (CNR) was calculated. Results Images reconstructed with IMR showed better detectability of PE than images reconstructed with FBP, especially at lower dose levels. With IMR, sensitivity was over 95% for central and segmental PE down to a dose level of 25%. Significantly higher subjective image quality was shown at lower dose levels (25% and 12.5%) for IMR images whereas it was higher for FBP images at higher dose levels. FBP was rated as showing less artificial image appearance. CNR was significantly higher with IMR at all dose levels. Conclusion By using IMR, a dose reduction of up to 50% while maintaining satisfactory image quality seems feasible in standard clinical situations, resulting in a mean effective dose of 1.38 mSv for CTPA.

Download Full-text

Deep learning versus iterative image reconstruction algorithm for head CT in trauma

Emergency Radiology ◽

10.1007/s10140-021-02012-2 ◽

2022 ◽

Author(s):

Zlatan Alagic ◽

Jacqueline Diaz Cardenas ◽

Kolbeinn Halldorsson ◽

Vitali Grozman ◽

Stig Wallgren ◽

...

Keyword(s):

Deep Learning ◽

Image Reconstruction ◽

Image Quality ◽

Reconstruction Algorithm ◽

Weighted Kappa ◽

Trauma Patients ◽

Reconstruction Algorithms ◽

Head Ct ◽

Image Reconstruction Algorithm ◽

Experienced Reader

Abstract Purpose To compare the image quality between a deep learning–based image reconstruction algorithm (DLIR) and an adaptive statistical iterative reconstruction algorithm (ASiR-V) in noncontrast trauma head CT. Methods Head CT scans from 94 consecutive trauma patients were included. Images were reconstructed with ASiR-V 50% and the DLIR strengths: low (DLIR-L), medium (DLIR-M), and high (DLIR-H). The image quality was assessed quantitatively and qualitatively and compared between the different reconstruction algorithms. Inter-reader agreement was assessed by weighted kappa. Results DLIR-M and DLIR-H demonstrated lower image noise (p < 0.001 for all pairwise comparisons), higher SNR of up to 82.9% (p < 0.001), and higher CNR of up to 53.3% (p < 0.001) compared to ASiR-V. DLIR-H outperformed other DLIR strengths (p ranging from < 0.001 to 0.016). DLIR-M outperformed DLIR-L (p < 0.001) and ASiR-V (p < 0.001). The distribution of reader scores for DLIR-M and DLIR-H shifted towards higher scores compared to DLIR-L and ASiR-V. There was a tendency towards higher scores with increasing DLIR strengths. There were fewer non-diagnostic CT series for DLIR-M and DLIR-H compared to ASiR-V and DLIR-L. No images were graded as non-diagnostic for DLIR-H regarding intracranial hemorrhage. The inter-reader agreement was fair-good between the second most and the less experienced reader, poor-moderate between the most and the less experienced reader, and poor-fair between the most and the second most experienced reader. Conclusion The image quality of trauma head CT series reconstructed with DLIR outperformed those reconstructed with ASiR-V. In particular, DLIR-M and DLIR-H demonstrated significantly improved image quality and fewer non-diagnostic images. The improvement in qualitative image quality was greater for the second most and the less experienced readers compared to the most experienced reader.

Download Full-text