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

Abstract

Background: New reconstruction algorithms based on deep learning have been developed to correct the image texture changes related to the use of iterative reconstruction algorithms. The purpose of this study was to evaluate the impact of a new deep learning image reconstruction [Advanced intelligent Clear-IQ Engine (AiCE)] algorithm on image-quality and dose reduction compared to a hybrid iterative reconstruction (AIDR 3D) algorithm and a model-based iterative reconstruction (FIRST) algorithm.Methods: Acquisitions were carried out using the ACR 464 phantom (and its body ring) at six dose levels (volume computed tomography dose index 15/10/7.5/5/2.5/1 mGy). Raw data were reconstructed using three levels (Mild/Standard/Strong) of AIDR 3D, of FIRST and AiCE. Noise-power-spectrum (NPS) and task-based transfer function (TTF) were computed. Detectability index was computed to model the detection of a small calcification (1.5-mm diameter and 500 HU) and a large mass in the liver (25-mm diameter and 120 HU).Results: NPS peaks were lower with AiCE than with AIDR 3D (−41%±6% for all levels) or FIRST (−15%±6% for Strong level and −41%±11% for both other levels). The average NPS spatial frequency was lower with AICE than AIDR 3D (−9%±2% using Mild and −3%±2% using Strong) but higher than FIRST for Standard (6%±3%) and Strong (25%±3%) levels. For acrylic insert, values of TTF at 50 percent were higher with AICE than AIDR 3D and FIRST, except for Mild level (−6%±6% and −13%±3%, respectively).For bone insert, values of TTF at 50 percent were higher with AICE than AIDR 3D but lower than FIRST (−19%±14%). For both simulated lesions, detectability index values were higher with AICE than AIDR 3D and FIRST (except for Strong level and for the small feature; −21%±14%). Using the Standard level, dose could be reduced by −79% for the small calcification and −57% for the large mass using AICE compared to AIDR 3D. Conclusions:The new deep learning image reconstruction algorithm AiCE generates an image-quality with less noise and/or less smudged/smooth images and a higher detectability than the AIDR 3D or FIRST algorithms. The outcomes of our phantom study suggest a good potential of dose reduction using AiCE but it should be confirmed clinically in patients.

Unenhanced abdominal low-dose CT reconstructed with deep learning-based image reconstruction: image quality and anatomical structure depiction

“…To verify the utility of DLIR in LDCT protocol, several phantom studies had performed by comparing hybrid- or model based IR algorithms and showed that DLIR demonstrate less noise and higher detectability in same dose reduction level compared with the other reconstruction algorithms [ 20 , 21 ]. In our phantom study, the NPS value was higher in LDCT reconstructed with DLIR than in SDCT reconstructed with ASiR-V of 40% though spatial resolutions were almost same.…”

Computed Tomography 2.0

“…Detectability of simulated lung lesions was best with the smoothest level in DLR; a dose reduction potential of 81% to 94% was assumed. An overview of recently published articles on deep learning–based image reconstruction 5,9,47–87 is given in Table 5.…”

Image-quality characteristics in the longitudinal direction from different image-reconstruction algorithms during single-rotation volume acquisition on head computed tomography: A phantom study

“…The circular edge method using a Delrin ® cylinder sphere phantom was used to assess the TTF. [22][23][24][25] TTFs were measured on 29 consecutive axial images (table position: À70to+70 mm) using ImageJ software. In the first step, 360 concentration profiles (1 × 33 pixels) were acquired from the center of the Delrin ® cylinder cross-section across the Delrin ® edge in 1°increments (Figure 1(b)).…”

“…The NPS was measured at ROIs of 35 × 35 pixels at the center and four other locations using imQuest software (Duke University, Durham, 2018) (Figure 1(a)). [21][22][23] The NPS was measured from the mean value of 25 measurement points (five ROIs × five repeated scans) in each slice level and repeated on 29 consecutive axial images in the longitudinal direction. The NPS was obtained for FBP images and mild and strong HIR and MBIR images.…”