• AI
  • Molecular Imaging
  • CT
  • X-Ray
  • Ultrasound
  • MRI
  • Facility Management
  • Mammography

Deep Learning Reconstruction Offers High-Quality CT Images with Less Dose


This mathematical algorithm can reduce dose in pediatric CT scans by 52 percent.

Applying a deep learning reconstruction (DLR) algorithm to a pediatric CT scan can not only improve the image quality, but it can also reduce the radiation dose by more than half, a new study has revealed.

In an article published on Nov. 17 in Radiology, investigators from Cincinnati Children’s Hospital Medical Center demonstrated that DLR can outperform filtered back projection (FBP), statistical-based iterative reconstruction (SBIR), and model-based iterative reconstruction (MBIR) in providing CT images of high diagnostic quality.

“The use of deep learning reconstruction in the CT image reconstruction process is an innovative application of artificial intelligence,” said the team led by Samuel L. Brady, M.D., Ph.D., a medical physicist at Cincinnati Children’s, “and our results suggest that it has the potential to improve image quality, object detection accuracy, and radiologist confidence to potentially reduce patient radiation dose.”

The team examined a total of 19 pediatric CT scans captured between February 2018 and December 2018. They performed a comparison of object detectability for 15 objects with a diameter between 0.5 mm to 10 mm at four contrast difference levels – 50, 150, 250, and 350 HU. Three pediatric radiologists performed the reviews, assessing the azygos vein, right hepatic vein, common bile duct, and superior mesenteric artery. They looked at edge definition, quantum noise level, and object conspicuity, and gave images a score from 1-to-10, worst to best.

According to provider analysis and scores, DLR had a higher performance than other methods in a variety of ways. From what they found, CT DLR improved object detectability for objects smaller than 7 mm (50 HU), 4 mm (150 HU), 3 mm (250 HU), and 2.5 mm (350 HU). When compared to FBP, SBIR, and MBIR, DLR improved object detectability by 51 percent, 18 percent, and 11 percent, respectively.

In addition, DLR reduced image noise without the noise texture effects seen with MBIR, the team said, and it demonstrated a 52-percent greater dose reduction – which translates to 53-percent organ-specific reductions – over SBIR. Overall, they said, the reviewing radiologists preferred DLR, giving it a higher score that the other methods. Within one point, they rated full field-of-view images from DLR a 7, MBIR a 6.2, SBIR a 6.2, and FBP a 4.6.

Ultimately, the team said, DLR can be used to create images that offer clinical benefit.

“The DLR algorithm improved image quality and dose reduction without sacrificing noise texture and spatial resolution,” they said.

For more coverage based on industry expert insights and research, subscribe to the Diagnostic Imaging e-Newsletter here.

Related Videos
Can Fiber Optic RealShape (FORS) Technology Provide a Viable Alternative to X-Rays for Aortic Procedures?
Does Initial CCTA Provide the Best Assessment of Stable Chest Pain?
Nina Kottler, MD, MS
Practical Insights on CT and MRI Neuroimaging and Reporting for Stroke Patients
Related Content
© 2024 MJH Life Sciences

All rights reserved.