Euromedim 2006 :1st European Conference on Molecular Imaging Technology

CONTRAST NOISE BEHAVIOUR FOR A ROTATING SLAT COLLIMATED GAMMA CAMERA

9 May 2006, 14:00

1h

Palais du Pharo, Marseille

poster • System simulation, design and implementation Poster Session :Simulation, Modeling, Reconstruction

Speaker

Mr Roel Van Holen (Ghent University ELIS-MEDISIP)

Description

INTRODUCTION Traditional gamma camera imaging with a parallel hole collimated detector (PH) is limited by the sensitivity versus spatial resolution tradeoff which is intrinsic to the design. This results in a reduced contrast for small lesions. The higher geometrical sensitivity of a rotating slat (RS) collimated strip detector could improve the image quality considerably. By means of phantom measurements, this study investigates if a RS gamma camera, with a spatial resolution of 5 mm at 10 cm collimator distance, is able to gain an improved contrast/noise ratio compared to a traditional camera, equipped with a LEHR collimator. METHODS Projections measured with a RS collimated gamma camera are planar integrals of the activity distribution and contain less information compared to ray projections, acquired with a traditional parallel hole collimated gamma camera. Therefore the RS collimated detecor has to spin around its own axis in order to collect complete data. Since planar projections are collected at different spin angles, image reconstruction comparable to a regular SPECT reconstruction is needed to obtain planar images. In this study we use a Monte Carlo based model of the acquisition physics. Incorporation of this model in the forward and backward projection of the MLEM algorithm yielded a Monte Carlo based reconstruction technique (MLEM-MC). A 17 cm diameter disc of uniform activity was printed on a sheet of paper together with 12 hot lesions located inside the uniform disc. The lesions had diameters ranging from 4 to 20mm and a contrast of 4:1. The sheet was placed at 10 cm from the collimator and 10 different (400 seconds) acquisitions were performed on both the PH and the RS camera. The data from the RS device were reconstructed in 250 iterations using MLEM-MC. The images coming from the PH camera were interpolated to have the same pixel size as the reconstructed RS images (1,8x1,8mm). The contrast recovery coefficient (CRC), defined as the ratio of the lesion activity and the background activity, was calculated for each lesion over the 10 different realisations of the measurement. Afterwards, the mean CRC (mCRC) over all lesions was calculated. For each pixel, the pixel noise was calculated as the standard deviation divided by the mean value over the 10 realisations. Averaging over all pixels yielded a global noise level (NL) of a measurement. RESULTS With the NL matched, the RS image reconstructed with MLEM-MC (60 iterations) reached a 21,0% higher CRC for the smallest hot spot, a 7,6% higher CRC for the 20mm hot lesion and a mCRC that was 20,0% higher compared to the PH values. For an equal mCRC, the NL of the RS image was 40,3% lower compared to the PH NL. For obtaining the same NL on a parallel hole system, the measurement takes 2,81 times longer. CONCLUSIONS When the same imaging time is used for all acquisitions, a lower NL is obtained for RS images at the same mCRC. On the other hand, a higher mCRC can be obtained when looking at images of the same NL. For comparable image quality on both modalities, the imaging time could be significantly reduced using the RS system.