RSNA 2012 

Abstract Archives of the RSNA, 2012


LL-PHS-WE1C

Low-Dose Abdomen CT Images Reconstructed with Model-based Iterative Image Reconstruction (MBIR) and Adaptive Statistical Iterative Reconstruction (ASIR): Influence of Reformation Plane and Image Thickness on Image Quality and Radiologist Preference

Scientific Informal (Poster) Presentations

Presented on November 28, 2012
Presented as part of LL-PHS-WEPM: Physics Afternoon CME Posters

Participants

Ravi K. Kaza MD, Presenter: Nothing to Disclose
Joel Franklin Platt MD, Abstract Co-Author: Nothing to Disclose
Mahmoud Mouhamad al-Hawary MD, Abstract Co-Author: Nothing to Disclose
Amit Pandya MD, Abstract Co-Author: Nothing to Disclose

PURPOSE

To evaluate the perceived image quality and reader preference of low dose abdomen CT images reconstructed with MBIR in axial plane and coronal plane as compared to images reconstructed with ASIR

METHOD AND MATERIALS

30 low dose CT enterography examinations were reconstructed with MBIR and ASIR in axial 0.625 mm and 2.5 mm thickness and coronal 2 mm thickness. In a random order, two radiologists evaluated 90 sets of MBIR and ASIR images matched for slice thickness and graded the type of image preferred for evaluation of 1) bowel and mesentery and 2) liver and solid organs on a scale of 1 to 5 (1: Strongly prefer ASIR 2: Prefer ASIR 3: Neutral 4: Prefer MBIR 5: Strongly prefer MBIR). MBIR images were also graded for perceived artifacts of 1) Plasticky/pixelated appearance and 2) Jagged margins of vessels and organs on a scale of 1 to 3 (1: Prominently seen 2: Minimally seen 3: None appreciated). A third radiologist measured image noise as average of the standard deviation of HU of a region of interest in portal vein and spinal muscle.

RESULTS

The mean patient weight, CT Dose Index vol, dose length product and effective dose were 162.2lbs, 3.3 mGy, 181.5mGy.cm and 2.72 mSv respectively. The mean image preference score for bowel and mesentery on 0.625 mm axial, 2.5 mm axial, 2 mm coronal sets was 2.6, 2.8, 4.3 and for liver and solid organs was 2.4, 2.7, 4.3 respectively. There was a statistically significant increase (p<0.001) in image preference score for coronal MBIR images as compared to axial MBIR images. Mean score for perceived artifacts of pixelated appearance on 0.625 mm axial, 2.5 mm axial, 2 mm coronal MBIR images was 1.1, 1.1, 2 and for jagged margins of vessels was 1.1, 1.1, 2 respectively. Image noise of .625 mm axial ASIR and MBIR images was 49.3 and 19.1 respectively and on 2.5 mm axial images was 30.9 and 15.1 respectively (p<.001).

CONCLUSION

Reader assessment of image quality of MBIR images is better for coronal images as compared to axial images. Perception of MBIR image artifacts is decreased in the coronal plane.

CLINICAL RELEVANCE/APPLICATION

Primary review in coronal plane could lead to improved acceptance of image quality of low dose MBIR images of abdomen.

Cite This Abstract

Kaza, R, Platt, J, al-Hawary, M, Pandya, A, Low-Dose Abdomen CT Images Reconstructed with Model-based Iterative Image Reconstruction (MBIR) and Adaptive Statistical Iterative Reconstruction (ASIR): Influence of Reformation Plane and Image Thickness on Image Quality and Radiologist Preference.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12043855.html