An Information-Theoretical Approach to Image Resolution Applied to Neutron Imaging Detectors based upon Individual Discriminator Signals

TL;DR

This paper introduces an information-theoretical measure of image resolution for neutron detectors, using mutual information to evaluate and compare the effectiveness of various signal processing algorithms in resolving closely spaced neutron impacts.

Contribution

It proposes a novel, calibration-independent resolution metric based on mutual information, and applies it to assess different algorithms and detectors in neutron imaging.

Findings

Center-of-gravity methods improve resolution over standard wire algorithms

The mutual information-based measure effectively quantifies resolution performance

Different detectors show varying resolution capabilities when evaluated with this method

Abstract

1D or 2D neutron imaging detectors with individual wire or strip readout using discriminators have the advantage of being able to treat several neutron impacts partially overlapping in time, hence reducing global dead time. A single neutron impact usually gives rise to several discriminator signals. In this paper, we introduce an information-theoretical definition of image resolution. Two point-like spots of neutron impacts with a given distance between them act as a source of information (each neutron hit belongs to one spot or the other), and the detector plus signal treatment is regarded as an imperfect communication channel that transmits this information. The maximal mutual information obtained from this channel as a function of the distance between the spots allows to define a calibration-independent measure of resolution. We then apply this measure to quantify the power of resolution of different algorithms treating these individual discriminator signals which can be implemented in firmware. The method is then applied to different detectors existing at the ILL. Center-of-gravity methods usually improve the resolution over best-wire algorithms which are the standard way of treating these signals.