Deep Learning Versus Iterative Reconstruction for CT Pulmonary Angiography in the Emergency Setting: Improved Image Quality and Reduced Radiation Dose

Abstract

To compare image quality and the radiation dose of computed tomography pulmonary angiography (CTPA) subjected to the first deep learning-based image reconstruction (DLR) (50%) algorithm, with images subjected to the hybrid-iterative reconstruction (IR) technique (50%). One hundred forty patients who underwent CTPA for suspected pulmonary embolism (PE) between 2018 and 2019 were retrospectively reviewed. Image quality was assessed quantitatively (image noise, signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR)) and qualitatively (on a 5-point scale). Radiation dose parameters (CT dose index, CTDIvol; and dose-length product, DLP) were also recorded. Ninety-three patients were finally analyzed, 48 with hybrid-IR and 45 with DLR images. The image noise was significantly lower and the SNR (24.4 ± 5.9 vs. 20.7 ± 6.1) and CNR (21.8 ± 5.8 vs. 18.6 ± 6.0) were significantly higher on DLR than hybrid-IR images (p < 0.01). DLR images received a significantly higher score than hybrid-IR images for image quality, with both soft (4.4 ± 0.7 vs. 3.8 ± 0.8) and lung (4.1 ± 0.7 vs. 3.6 ± 0.9) filters (p < 0.01). No difference in diagnostic confidence level for PE between both techniques was found. CTDIvol (4.8 ± 1.4 vs. 4.0 ± 1.2 mGy) and DLP (157.9 ± 44.9 vs. 130.8 ± 41.2 mGy∙cm) were lower on DLR than hybrid-IR images. DLR both significantly improved the image quality and reduced the radiation dose of CTPA examinations as compared to the hybrid-IR technique.

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

“…However, many studies have shown that the potential dose reduction obtained with IR algorithms was limited to low contrast liver lesions and the outcomes found on phantoms overestimated the real dose reduction in patients (44)(45)(46). Nevertheless, the first patient studies published on pulmonary or cardiac CT angiography and chest and abdominal CT have also confirmed that AiCE improved both image quality and lesion detection as compared with AIDR 3D for a given dose level (22,25,26,30) or with a dose reduction (26,29,34,35). Singh et al found that the dose could be reduced by −84% between a standard protocol (CTDI vol : 13.0±4.4 mGy) with AIDR 3D and a low-dose protocol (CTDI vol : 2.1±0.8 mGy) with AiCE for the detection of the same abdominal lesions and an overall image quality scored acceptable for more than 95% of patients (34).…”

“…In addition, the TrueFidelity DNN was trained with both patient and phantom data whereas it was trained only with patient data for AiCE. CT images obtained with these algorithms using denoising techniques showed suppressed noise with no change of noise texture or distortion of anatomical and pathological structures (19,(22)(23)(24)(25)(26)(27)(28)(29)(30).…”

“…In addition, in that same study, only the iterative level "Standard" was assessed and not the three levels available for each IR algorithms (Mild, Standard and Strong). On another note, many clinical studies have shown the impact of AiCE on dose reduction and image quality for different anatomical locations in both adult and pediatric patients (22,(25)(26)(27)29,34,35).…”

Experimental Analysis of the Utility of Liquid Metal Gallium as a Contrast Agent for CT Hepatic Artery Angiography in Living Rabbits

“…CT difference between the hepatic artery and liver parenchyma in the arterial phase; CT difference between the abdominal aorta and the liver parenchyma in the arterial phase. Contrast-to-noise ratio (CNR) of the hepatic artery and abdominal aorta in the arterial phase, CNR=(CT (target vessel)-CT (same layer of the liver parenchyma))/SD (same layer of the liver parenchyma) 39 .…”

Assessment of Radiation Dose in Medical Imaging and Interventional Radiology Procedures for Patient and Staff Safety

“…During the process of the optimization of protection in medical imaging, the use of modern imaging systems and efficient reconstruction techniques is shown to be effective in order to reduce patient radiation dose. A deep learning reconstruction significantly reduced image noise and improved overall image quality of CT pulmonary angiography examinations in the emergency setting and offered an additional significant radiation dose reduction while allowing slices to be twice as thin as compared to hybrid iterative reconstruction (IR) [ 2 ]. The use of an Advanced Modeled IR (ADMIRE; Siemens Healthineers, Forchheim, Germany) allowed a reduction in the radiation dose to 25% of the original dose with the same diagnostic accuracy for the assessment of neck abscesses [ 3 ].…”