Validation of compound Poisson noise model for computed tomography with energy-integrating detector

These instructions provide guidelines for preparing manuscripts for submission to the Conference Record (CR) of the 2012 IEEE Nuclear Science Symposium and Medical Imaging Conference. If you are using Microsoft Word 6.0 or later to prepare your manuscript, you should use this document as a template. Define all symbols used in the abstract. Do not cite references in the abstract. Traditional Poisson noise model underestimates noise for large object or patient sizes. We suggest a method for more accurate noise estimation (energy and object size dependent) by using a compound Poisson noise model. Our results with water phantom data show validity of the proposed noise model. Our results confirm more accurate noise estimation compared to a simple Poisson model. The new noise model has potential to improve noise accuracy in low dose simulations, data-domain noise reduction, and iterative reconstruction.