Comparison of two deep learning image reconstruction algorithms in chest CT images: A task-based image quality assessment on phantom data - 06/09/21

Doi : 10.1016/j.diii.2021.08.001

Joël Greffier ^a,^{^{1
Both authors contributed equally to this work.},}^⁎ , Julien Frandon ^a,^{^{1
Both authors contributed equally to this work.}}, Salim Si-Mohamed ^b, Djamel Dabli ^a, Aymeric Hamard ^a, Asmaa Belaouni ^a, Philippe Akessoul ^a, Francis Besse ^c, Boris Guiu ^d, Jean-Paul Beregi ^a
^a Department of Medical Imaging, CHU Nimes, Univ Montpellier, Medical Imaging Group Nimes, EA 2992, 30029 Nîmes, France
^b Department of Radiology, Hospices Civils de Lyon, 69500 Lyon, France
^c Department of Radiology Centre Cardiologique Nord, 93200 Saint Denis, France
^d Department of Radiology Saint-Eloi University Hospital, 34295 Montpellier, France

^⁎Corresponding author.

Sous presse. Épreuves corrigées par l'auteur. Disponible en ligne depuis le Monday 06 September 2021
Cet article a été publié dans un numéro de la revue, cliquez ici pour y accéder

Highlights

•	Using TrueFidelity^TM algorithm, spatial resolution is independent of dose and contrast and image texture modification is limited.
•	Using AiCE algorithm, spatial resolution depends on dose and contrast and image is smoother, especially at highest level.
•	Detectability of chest lesions is greater with AiCE than with TrueFidelity^TM at low dose levels but AiCE changes noise texture.

Le texte complet de cet article est disponible en PDF.

Abstract

Purpose

The purpose of this study was to compare the effect of two deep learning image reconstruction (DLR) algorithms in chest computed tomography (CT) with different clinical indications.

Material and methods

Acquisitions on image quality and anthropomorphic phantoms were performed at six dose levels (CTDI_vol: 10/7.5/5/2.5/1/0.5mGy) on two CT scanners equipped with two different DLR algorithms (TrueFidelity^TM and AiCE). Raw data were reconstructed using the filtered back-projection (FBP) and the lowest/intermediate/highest DLR levels (L-DLR/M-DLR/H-DLR) of each algorithm. Noise power spectrum, task-based transfer function (TTF) and detectability index (d’) were computed: d’ modelled detection of a soft tissue mediastinal nodule, ground-glass opacity, or high-contrast pulmonary lesion. Subjective image quality of anthropomorphic phantom images was analyzed by two radiologists.

Results

For the L-DLR/M-DLR levels, the noise magnitude was lower with TrueFidelity^TM than with AiCE from 2.5 to 10 mGy. For H-DLR, noise magnitude was lower with AiCE . For L-DLR and M-DLR, the average NPS spatial frequency (f_av) values were greater for AiCE except for 0.5 mGy. For H-DLR levels, f_av was greater for TrueFidelity^TM than for AiCE. TTF_50% values were greater with AiCE for the air insert, and lower than TrueFidelity^TM for the polyethylene insert. From 2.5 to10 mGy, d’ was greater for AiCE than for TrueFidelity^TM for H-DLR for all lesions, but similar for L-DLR and M-DLR. Image quality was rated clinically appropriate for all levels of both algorithms, for dose from 2.5 to 10 mGy, except for L-DLR of AiCE.

Conclusion

DLR algorithms reduce the image-noise and improve lesion detectability. Their operations and properties impacted both noise-texture and spatial resolution.

Le texte complet de cet article est disponible en PDF.

Keywords : Multidetector computed tomography, Task-based image quality assessment, Deep learning image reconstruction

Abbreviations : AiCE, CT, DNN, DLR, GGO, HCP, HU, IR, NPS, ROI, SD, TF, TTF