Apparatus and process for optimizing radiation detection counting times using machine learning

#Systems used to detect radiation or determine the content of radioactive material are common in a multitude of nuclear science applications including, but not limited to, passive and active interrogation systems located at border crossings, common gamma spectroscopy systems used to analyze samples in a variety of environments in the nuclear industry, and detection of radiation from therapeutic treatments or diagnostic imaging applications in nuclear or imaging medicine. Most practical applications involve a radiation detector (of which there are many kinds), process signal electronics, and some analytical approaches to determining the presence and/or amount of radiation or radioactive material being measured. The analytical approaches typically used involve collection of detector data over a finite or defined period of time followed by analysis of the integrated counts collected over that time. The start-and-stop process is repeated and necessarily involves time periods between collection cycles where the analysis is being performed instead of collecting more data. Herein, a novel approach is described that utilizes machine-learning techniques to eliminate this cyclic pattern and provides information in real-time.

#Machine learning, which is also referred to herein as statistical analysis, data mining, or data analytics, is currently regarded as a subfield of computer science. However, the applicability of machine learning to tasks not commonly attributed to computer science is widely touted. Through algorithmic methods, machine learning is able to address complex data tasks. Scientists postulate the data creation or curation from a particular model or they assume an unknown model and apply algorithmic methods to the data to reach conclusions. The latter ensures a data-driven process and supports an innovative way of solving complex data problems. Machine learning techniques are increasingly used in the radiation space discipline-wide. For example, neural networks (NN) and support vector machines-based solutions such as interpreted gamma-ray spectroscopy data and a neutral network approach recognizing patterns for spectral analysis from High-purity Germanium (HPGe) detectors are known machine learning applications. However, these approaches are limited in their detection speed and post detection operations.

#The background description provided here is for the purpose of generally presenting the context of the disclosure. Work of the presently named inventors, to the extent it is described in this background section, as well as aspects of the description that may not otherwise qualify as prior art at the time of filing, are neither expressly nor impliedly admitted as prior art against the present disclosure. Unless otherwise indicated here, the approaches described in this section are not prior art to the claims in the present disclosure and are not admitted to be prior art by inclusion in this section.

#The embodiments of the disclosure will be understood more fully from the detailed description given below and from the accompanying drawings of various embodiments of the disclosure, which, however, should not be taken to limit the disclosure to the specific embodiments, but are for explanation and understanding only.

#FIG. 1A illustrates an apparatus and/or system with active emission and detection of radiation for optimizing nuclear or radiological counting using machine learning, in accordance with some embodiments.

#FIG. 1B illustrates an apparatus and/or system with passive detection of radiation for optimizing nuclear or radiological counting using machine learning, in accordance with some embodiments.

#Some embodiments describe an apparatus and method for performing an analytical method to minimize the time necessary to produce a required signal in an environment. Various embodiments here are described with reference to nuclear or radiological counting environments. However, the embodiments are not limited to such and can be used for any suitable environment where a material is to be identified from raw data and then a decision needs to be made about the usage of the material. By optimizing (e.g., minimizing or reducing) the counting time of the detected radioactive material, the speed and efficiency of determining the amount or presence of the radioactive material or obtaining the required contrast is improved. In some embodiments, a dynamic material identification counting architecture, called Frieder Counting, is provided. This counting architecture involves collecting raw count data from a counting system and using machine learning, data mining, or statistical or data analytics techniques, used interchangeably herein, to both optimize (e.g., minimizing or reducing) the counting times involved and to provide real-time indication of the presence or identification of radiation or radionuclides.

#Gamma spectroscopy is one example of how the method of various embodiments is applied. Gamma spectroscopy is the process of collecting photon interactions with a detector to determine information on the energy of the original photon. Detectors used in gamma spectroscopy involve materials where energy information can be collected. In one non-limiting example, NaI(Th) scintillators or solid state semiconductor detectors such as high purity germanium detectors are used.

#As is known in the art, for example, U.S. Pat. No. 7,711,661 issued to Gentile et al., such systems usually involve a series of electronics to perform signal processing. This process, illustrated in FIG. 1A and FIG. 1B, includes but is not limited to: (1) collecting pulses from a detector, (2) amplifying and conditioning the signal detected with a preamplifier, (3) shaping the signal with an amplifier, (4) using an analog-to-digital converter to digitize the signal, (5) using a multi-channel analyzer to associate the signal with a channel or energy, and eventually (6) entering, either directly electronically or manually, this aforementioned information into a software package. The software package serves as a platform for performing identification and/or quantification and as an interface with the user, in accordance with various embodiments.