Sachin Moghe PhD, Presenter: Nothing to Disclose

Bernice E. Hoppel PhD, Abstract Co-Author: Employee, Toshiba Corporation

Developing reduced dose protocols and novel dose reduction algorithms require validation at multiple dose levels. However, repeated scanning of clinical patients is often not possible. Raw data noise simulation allows creation of virtual low dose acquisitions without repeated scanning. This study will validate a custom noise simulation tool in a COPDgene phantom by comparing simulated low dose scans with actual reduced dose scans.

A COPDgene phantom was scanned using a 320-row CT scanner. Seven helical scans were acquired with decreasing mAs (100, 75, 50, 40, 25, 15 and 5mAs) representing doses at 100 %, 75%, 50%, 40%, 25%, 15%, and 5% of the typical clinical protocol, respectively. Scans were acquired at 120 kV, 0.5s rotation, and 64 x 0.5mm detector configuration. To simulate reduced dose acquisitions, noise was added to the raw data of the 100 mAs scan. The noise simulation tool creates a precise model of the scanner noise based on direct measurements from detector, DAS, and x-ray spectrum. Acquired (AD) and simulated (SimD) raw data were reconstructed with 1mm slices with the clinical FC03 kernel and beam hardening correction. Means and standard deviations (SD) were measured for both the AD and SimD cases in 7 ROIs placed in the phantom. Pixel noise was compared between AD and SimD using linear correlation.

The difference in mean HU for each of the 7 ROIs was less than 2.5 HU at all dose levels, except the 5% dose level, corresponding to an average error of less than 3%. The average SD difference in noise between the AD and SimD cases was 0.7(+/-0.3) HU over all ROIs and dose levels excluding the 5% level with a linear correlation of 1.02+/- 0.01 and r2 = 0.999. This corresponds to a deviation of 3-5% increasing to 8% at the 5% dose level. At the 5% dose level there was a maximum difference in the mean of 6.3 HU and an average(SD) deviation in noise of 1.9(+/-2.1).

Higher errors at 5mAs are attributable to reaching quantum noise level and photon starvation. The noise simulation tool shows excellent correlation compared with actual low dose scans allowing the user the capability to simulate dose reduction scenarios without repeated patient scans.

Accurate raw data noise simulation can aid in the development of low dose scan protocols and test novel reconstruction methods in clinical chest patients without requiring multiple acquisitions.

Moghe, S, Hoppel, B, Validation of Noise Simulation Tool Using the COPDgene Phantom: Comparing Acquired and Simulated Low-Dose CT Scans.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12043877.html