Waveform processing using neural network algorithms on the front-end electronics

In a multi-channel radiation detector readout system, waveform sampling, digitization, and raw data transmission to the data acquisition system constitute a conventional processing chain. The deposited energy on the sensor is estimated by extracting peak amplitudes, area under pulse envelopes from the raw data, and starting times of signals or time of arrivals. However, such quantities can be estimated using machine learning algorithms on the front-end Application-Specific Integrated Circuits (ASICs), often termed as “edge computing”. Edge computation offers enormous benefits, especially when the analytical forms are not fully known or the registered waveform suffers from noise and imperfections of practical implementations. In this work, we aim to predict peak amplitude from a single waveform snippet whose rising and falling edges containing only 3 to 4 samples. We thoroughly studied two well-accepted neural network algorithms, Multi-Layer Perceptron (MLP) and Convolutional Neural Network (CNN) by varying their model sizes. To better fit front-end electronics, neural network model reduction techniques, such as network pruning methods and variable-bit quantization approaches, were also studied. By combining pruning and quantization, our best performing model has the size of 1.5 KB, reduced from 16.6 KB of its full model counterpart. It can reach mean absolute error of 0.034 comparing to that of a naive baseline of 0.135. Such parameter-efficient and predictive neural network models established feasibility and practicality of their deployment on front-end ASICs.

Low-Temperature Annealing of CdZnTeSe under Bias

We performed a gradual low-temperature annealing up to 360 K on a CdZnTeSe radiation detector equipped with gold and indium electrodes under bias at both polarities. We observed significant changes in the detector’s resistance and space-charge accumulation. This could potentially lead to the control and improvement of the electronic properties of the detector because the changes are accompanied with the reduction in the bulk dark current and surface leakage current. In this article, we present the results of a detailed study of the internal electric field and conductivity changes in CdZnTeSe detector for various annealing steps under bias taking into account different polarities during annealing and subsequent characterization. We observed that low-temperature annealing results in an increase in the barrier height at the contacts that, in general, reduces the dark current and decreases the positive space charge present in the sample compared to the pre-annealed condition.

A design of forceps-type coincidence radiation detector for intraoperative LN diagnosis: clinical impact estimated from LNs data of 20 esophageal cancer patients

Purpose To reduce postoperative complications, intraoperative lymph node (LN) diagnosis with 18F-fluoro-2-deoxy-D-glucose (FDG) is expected to optimize the extent of LN dissection, leading to less invasive surgery. However, such a diagnostic device has not yet been realized. We proposed the concept of coincidence detection wherein a pair of scintillation crystals formed the head of the forceps. To estimate the clinical impact of this detector, we determined the cut-off value using FDG as a marker for intraoperative LN diagnosis in patients with esophageal cancer, the specifications needed for the detector, and its feasibility using numerical simulation. Methods We investigated the dataset including pathological diagnosis and radioactivity of 1073 LNs resected from 20 patients who underwent FDG-positron emission tomography followed by surgery for esophageal cancer on the same day. The specifications for the detector were determined assuming that it should measure 100 counts (less than 10% statistical error) or more within the intraoperative measurement time of 30 s. The detector sensitivity was estimated using GEANT4 simulation and the expected diagnostic ability was calculated. Results The cut-off value was 620 Bq for intraoperative LN diagnosis. The simulation study showed that the detector had a radiation detection sensitivity of 0.96%, which was better than the estimated specification needed for the detector. Among the 1035 non-metastatic LNs, 815 were below the cut-off value. Conclusion The forceps-type coincidence detector can provide sufficient sensitivity for intraoperative LN diagnosis. Approximately 80% of the prophylactic LN dissections in esophageal cancer can be avoided using this detector.

Quartz Tuning Fork Sensor-Based Dosimetry for Sensitive Detection of Gamma Radiation

This study generally relates to nuclear sensors and specifically to detecting nuclear and electromagnetic radiation using an ultrasensitive quartz tuning fork (QTF) sensor. We aim to detect low doses of gamma radiation with fast response time using QTF. Three different types of QTFs (uncoated and gold coated) were used in this study in order to investigate their sensitivity to gamma radiations. Our results show that a thick gold coating on QTF can enhance the quality factor and increase the resonance frequency from 32.7 to 32.9 kHz as compared to uncoated QTF. The results also show that increasing the surface area of the gold coating on the QTF can significantly enhance the sensitivity of the QTF to radiation. We investigated the properties of gold-coated and uncoated QTFs before and after irradiation by scanning electron microscopy. We further investigated the optical properties of SiO2 wafers (quartz) by spectroscopic ellipsometry (SE). The SE studies revealed that even a small change in the microstructure of the material caused by gamma radiation would have an impact on mechanical properties of QTF, resulting in a shift in resonance frequency. Overall, the results of the experiments demonstrated the feasibility of using QTF sensors as an easy to use, low-cost, and sensitive radiation detector.

DEVELOPMENT OF A SMART OCEAN RADIATION MONITORING SYSTEM

Ocean radiation monitoring systems (ORMSs) are an essential component in the radiation early warning network that monitors radiation exposure and estimates radioactive propagation induced by nuclear activities or nuclear accidents in the sea. Numerous systems have been developed and installed in the radiation warning network in different countries. However, there is not any similar product that has been studied and developed in Vietnam. This paper presents a complete process in designing and manufacturing a marine buoy integrated with a radiation sensor. The radiation detector can measure both dose rate and radiological spectrum. The ORMS also combines multimodal data transmission and various programmed software for data processing, signal transmission, and system control. Therefore, the proposed configuration system has potential application in terms of performance and maintenance.

Learning gamma-ray scanning technique through an educational simulation rig

Gamma-ray scanning is non-destructive testing (NDT) technique to inspect problems in distillation columns of industrial plant. In order to familiarize with the problems’ identification and data interpretation, an attempt has been made by constructing an educational rig for simulation of gamma-ray scanning technique in a laboratory scale. The rig is made of cylindrical transparent acrylic material with a diameter and a height of 0.3 m and 2.4 m, respectively. The rig consists of artificial trays and packed bed structures designed in such a way that it represents a duplicate of the distillation unit but in a miniature size. The rig is also equipped with a submerged water pump and water tank which both are located at the bottom part of the rig. Water from the tank are circulated by a water pump through a pvc tube to a distributor at the upper part of the rig. In the rig, water is flowing downward gravitically for passing through packed bed and trays structures until terminated at the tank. The gamma-ray scanning experiment was carried out by moving the radiation detector and Cs-37 gamma-ray source simultaneously from top to bottom for every 5 cm moving step. Artificial problems such as flooding, tray positions, collapsed trays are clearly identified. The lesson learned from this experiment concludes that simulation of gamma ray scanning is very suitable for troubleshooting and diagnosing malfunctions of the internal structure of the distillation column.

High-Resolution Distributed Radiation Detector System Assisted by Intelligent Image Recognition

With the development of machine learning and image recognition technology, the detector system tends to be standardized and intelligent. However, large numbers of distributed radiation detectors connected to the power grid will bring huge uncertainty to the operation of the power grid and even cause certain interference. The monitoring system of the distributed radiation detectors can control the running status of the distributed photoelectric detection system in real-time and guarantee the safe and stable operation of the detector system. This article proposes an improved genetic detector system to avoid “blind spots” in the radiation detector monitoring based on the characteristics of photovoltaic (PV) arrays, which are considered as individual pixels, and then the reliability of the monitoring is ensured by the monitoring coverage of these pixels by the detector nodes. The performance of the radiation detector monitoring is restored by activating those spare nodes with sufficient energy to replace those that fail, ensuring that the distributed detection system can be monitored in a timely and efficient manner at all times. The simulation results confirm the reasonable validity of the algorithm.