The potential for reduced radiation dose from deep learning-based CT image reconstruction

Abstract

The purpose of this phantom study is to compare radiation dose and image quality of abdominal computed tomography (CT) scanned with different tube voltages and tube currents, reconstructed with filtered back projection (FBP), hybrid iterative reconstruction (IR) and deep learning image reconstruction (DLIR) algorithms. A total of 15 CT scans of whole body phantoms were taken with 3 different tube voltages and 5 different tube currents. The images were reconstructed with FBP, 30% and 50% hybrid IR adaptive statistical iterative reconstruction (ASIR-V), and low, medium and high strength DLIR algorithms. The image scanned with tube voltage/tube current of 120 kV/ 200 mA and reconstructed with FBP algorithm was chosen as the reference image. Five radiologists independently analyzed the images individually and also compared it with the reference image in overall, using the visual grading analysis. The mean score of each image was calculated and compared. Using DLIR algorithms, the radiation dose was reduced by 65.5% to 68.1% compared with the dose used in the reference image, while maintaining comparable image quality. Using the DLIR algorithm of medium strength, the image quality was even better than the reference image with a reduced radiation dose up to 36.2% to 50.0%. The DLIR algorithms generated better quality images than ASIR-V algorithms in all the data sets. In addition, among the data sets reconstructed with DLIR algorithms, image quality was the best at the medium strength level, followed by low and high. This phantom study suggests that DLIR algorithms may be considered as a new reconstruction technique by reducing radiation dose while maintaining the image quality of abdominal CTs.

First Performance Evaluation of an Artificial Intelligence-Based Computer-Aided Detection System for Pulmonary Nodule Evaluation in Dual-Source Photon-Counting Detector CT at Different Low-Dose Levels

“…[32][33][34] Therefore, it is hoped that the negative aspects of iterative image reconstruction in EID-CT can be overcome with the use of AI-based image reconstruction techniques in the future. [35][36][37] Overall, the AI-CAD system proved to be an applicable tool in nodule detection in PCD-CT showing comparable sensitivity for pulmonary nodule detection to EID-CT with the HR-mode. As expected, the false-positive rate increased at lower radiation dose levels.…”

“…Artificial intelligence-based image reconstruction allows image noise suppression without changing noise texture or affecting anatomical and pathological structures 32–34 . Therefore, it is hoped that the negative aspects of iterative image reconstruction in EID-CT can be overcome with the use of AI-based image reconstruction techniques in the future 35–37 …”

Radiation Dose Reduction Opportunities in Vascular Imaging

“…Convolutional neural networks (CNN) are AI algorithms developed to recreate standard-dose images from low-dose computed tomography (LDCT) scans [ 26 ]. Deep learning image reconstruction (DLIR) uses neural networks to develop images with lower noise compared to FBP and ASIR images using 60% less radiation [ 27 ]. With a novel, modular approach that allows radiologists to view improvements in image quality iteratively, deep learning (DL) algorithms can improve the image quality of LDCT across multiple different vendors, with the resulting image quality rivaling normal dose computed tomography (NDCT) scans using traditional iterative reconstruction methods [ 28 ].…”

Enhanced Visualization in Endoleak Detection: A Comparative Analysis of Virtual Monoenergetic Imaging and Advanced Reconstruction Techniques

“…Few studies have been conducted on the performance of DLM algorithms for cardiac and vascular CT 32,[56][57][58][59] . In cardiac CT, vessel contouring is adequate with routine-dose FBP, IR, and MBIR, but deteriorates with low-dose scans due to increased noise and reduced spatial resolution.…”