Program Information
Analytical Calculation of Scatter Projections in Nuclear Medicine Imaging
F Kalantari1*, H Rajabi2 , A Rodriguez1 , s gholami3 , M Tavakoli4 , J Wang1 , (1) UT Southwestern Medical Center, Dallas, TX, (2) Tarbiat Modares University, Tehran, Tehran, (3) Cancer institute, Tehran, Tehran, (4) Indiana University-Purdue University, Indianapolis, Indiana
Presentations
WE-F-201-9 (Wednesday, August 2, 2017) 1:45 PM - 3:45 PM Room: 201
Purpose: Scattered photons degrade the image quality and quantitative accuracy of SPECT and PET images. Dual window (DW) and triple energy window (TEW) methods are conventional methods of scatter estimation. They incorrectly assume that the scatter distribution in a photo-peak is similar to those from adjacent windows. Monte Carlo (MC) based scatter estimation and its incorporation into iterative Expectation Maximization methods offers the most accurate scatter correction. However, MC is very time consuming which limits its application for scatter correction. We introduced a force detection method for fast and accurate scatter modeling.
Methods: One million photons were tracked from their point of emission to the first interaction site inside the object to calculate a scatter map. This scatter map is then properly weighted depending on the detector positions and projected analytically to generate scatter projections. Two different weights are considered for each point within the scatter map. The first is the probability of scattering in a specific direction to be detected and is calculated based on the Klein-Nishina (KN) formula. The second is the detection probability of photons in a photo-peak after losing energy from Compton scattering. Both weighting factors vary by scattering angle and are detector position dependent with the final weight being the product of two independent probabilities. Our calculation was repeated for typical nuclear medicine isotope energies. Scatter projections of an XCAT digital phantom were created for 60 different angles and compared to those from SIMIND Monte Carlo simulator as reference.
Results: The RMS error compared to the reference scatter projections were 3.43±0.46 in our method. The corresponding values were 6.46±0.83 in DW and 4.44±0.47 in TW. Using our proposed method, 60 scatter projections were simulated in less than a minute.
Conclusion: We introduced a fast and accurate method for generating scatter projections in nuclear medicine imaging.
Contact Email: