Spotlight
Topics
DI Executive
Subscribe
- AI
- Molecular Imaging
- CT
- X-Ray
- Ultrasound
- MRI
- Facility Management
- Mammography
AI-Based Denoising for Neck CT May Facilitate Reductions in Radiation Dosing
Image quality, sharpness, and contrast with AI-based denoising were significantly enhanced for neck CT in comparison to conventional CT image reconstruction at 100 percent and 50 percent mAs, according to newly published research.
New research demonstrates that an artificial intelligence (AI) denoising software facilitates higher contrast-to-noise ratios (CNRs) and higher subjective assessments of image quality for head computed tomography (CT) scans than conventional CT at various radiation dosing levels.
For the retrospective study, published recently in the European Journal of Radiology, researchers examined the use of a deep learning denoising algorithm (ClariAce version 1.0.2, ClariPi) in 50 patients who had CT scans for suspected neck tumors. In addition to objective measures such as CNR, the study authors reviewed subjective assessments from three radiologists of image quality, sharpness, and contrast at 100 percent radiation and simulated dosing of 50 and 25 percent, according to the study.
For subjective image quality, the researchers found that the average mean assessment at 100 percent radiation dosing with conventional iterative CT reconstruction was 0.12 versus 0.65 for the denoising algorithm. At simulated 50 percent dosing, the average mean image quality was -0.43 for conventional CT reconstruction in comparison to 0.36 for the denoising algorithm.
Here one can see conventional neck CT images and the use of AI-based denoising at different radiation doses. (Images courtesy of the European Journal of Radiology.)
For the subjective assessments of image sharpness, the average mean rating at 100 percent dosing for conventional CT reconstruction was 0.11 in contrast to 0.67 for AI-enabled denoising, according to the study authors. At 50 percent radiation dosing, the researchers found the AI denoising algorithm had an average mean rating for sharpness of 0.39 in comparison to -0.47 for conventional CT reconstruction.
“In subjective analysis, image quality, diagnostic confidence, sharpness, and contrast were significantly higher for (the AI denoising algorithm) than for (conventional CT reconstruction) at 100 and 50% (dosing),” noted lead study author David Plajer, M.D., who is affiliated with the Department of Diagnostic and Interventional Radiology at Eberhard-Karls University in Tubingen, Germany, and colleagues.
In their evaluation of subjective assessments for image quality, the researchers found the average mean for conventional image reconstruction at 100 percent dosing was 0.12 in comparison to 0.07 with the AI denoising algorithm at a simulated 25 percent dosing. They noted similarities with other subjective measures such as contrast with an average mean of 0.19 for conventional image reconstruction at 100 percent dosing in contrast to 0.10 at 25 percent dosing for the AI algorithm.
“Remarkably, no significant differences existed between the subjective image quality characteristics of the 25 % mAs denoised and 100 % mAs (conventional CT reconstruction) data sets,” pointed out Plajer and colleagues.
The study authors also pointed out that the AI denoising algorithm had consistently higher CNRs than conventional CT reconstruction regardless of the radiation dosing level.
Beyond the inherent limitations of a retrospective study, the authors acknowledged the small sample size and low-dose simulations. The researchers also conceded potential bias with the study’s emphasis on subjective assessments of image quality and noted that extrapolation of the study findings may be limited due to the use of CT scans drawn from a single scanner.
Recent Videos
Related Content
FDA Clears AI-Powered Ultrasound Software for Cardiac Amyloidosis Detection
November 20th 2024Article
The AI-enabled EchoGo® Amyloidosis software for echocardiography has reportedly demonstrated an 84.5 percent sensitivity rate for diagnosing cardiac amyloidosis in heart failure patients 65 years of age and older.
