MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Correction to the Ground Truth Noise From Adjacent Slice Subtractions with the First and Third Generation Adaptive Statistical Iterative Reconstructions

Y (Jimmy) Zhou*, A Scott , Cedars-Sinai Medical Center, Los Angeles, CA

Presentations

(Monday, 7/15/2019) 1:15 PM - 1:45 PM

Room: Exhibit Hall | Forum 9

Purpose: Due to the unrealistic nature of multiple repeated acquisitions in practice, it is natural to assess the noise from adjacent slice subtractions. However, the result may not be the true noise due to the possible correlations. We attempted to correct the slice subtraction results to the ground truth noise at different spatial scales with different pitches and ASIR reconstructions.

Methods: A water phantom with 32-cm in diameter was scanned using a GE HD 750 CT with two pitches (0.516 and 1.375) at CTDIvol of 10 mGy. The scan was repeated ten times with the scan length of 230 mm. The first and third generation adaptive statistical reconstructions (ASIR and ASIR-V, both with 10%, 50% and 100%) were applied to reconstruct images with 5 mm thickness. To quantify the noise at various scales in either subtractions, a square region of 126x126 mm was partitioned to matrices of six element sizes (2 – 10 mm). For each spatial scale (element size), the mean pixel value distribution was computed from the matrix and the standard deviation was served as the noise. With a correction of 1/√2, average results from subtractions between the repeated scans were used as the ground truth noises. Similar method was used to obtain the results from the subtractions between adjacent slices. The results were compared to the ground true noise and the necessary correction was identified.

Results: The true noises were found to be 1.2 (± 0.047) times higher than the results from adjacent slice subtractions regardless of the pitch, spatial scale, and the fraction of ASIR or ASIR-V.

Conclusion: The noise obtained from adjacent slice subtraction needs to be corrected by a factor of 1.2. The finding can help establish the criteria for low contrast detectability based on the noise assessment.

Keywords

CT, Noise, Image Correlation

Taxonomy

IM- CT: Phantoms - physical

Contact Email