Virtual extrapolation technique for retracing line of response of single scattered events in positron emission tomography

TL;DR

This paper introduces a novel mathematical Virtual extrapolation technique to trace and incorporate single-scattered events in PET, aiming to enhance imaging sensitivity despite finite detector resolutions.

Contribution

The paper proposes a new method based on probability density functions to retrace scattered photon paths in PET, transforming data for improved event analysis.

Findings

Highest counts at real scattering points confirm model accuracy

Performance is optimal with ideal timing and energy resolution

Finite resolutions reduce the effectiveness of the method

Abstract

Purpose: The scattering phenomenon creates degrading effects in positron emission tomography (PET) and the corresponding events are rejected conventionally. We have proposed a mathematical model to retrace the original line of response of the single-scattered coincident events with the aim to incorporate such events in PET. Methods: We have devised a new Virtual extrapolation technique based on the concept of the probability density functions. Through which we transformed the original two-parameter list mode data for the coincident photon pairs to a one-parameter data set. The procedure of random sampling and sampling distribution - by utilizing some unique properties like a collective difference and the length compensation - was employed in the data analysis. We studied the effect of finite timing and energy resolution of detectors on the performance of the proposed model. Results: We determined the frequency of occurrence of the data values corresponding to real as well as fictitious scattering points { placed on the arc of a circle drawn with the constant scattering angle - to observe the highest counts. As expected, we found the highest counts at the location of real scattering points. The model was evaluated for a uniform attenuating phantom medium. The results were found impressive for the case of ideal time and energy information, less so for the cases of finite resolutions. Conclusions: Our work outlines a completely new approach; though, a lot more sophistication is required for its practical utility. Nonetheless, the technique seems promising in developing a potential approach to improve the sensitivity of PET imaging.