Signal Processing and Analysis Techniques for Nuclear Quadrupole Resonance Spectroscopy

Foreword
Preface
Contents
Abbreviations
1 Introduction
	1.1 Context of This Work
	1.2 Purpose of This Work
	1.3 Structure of This Work
	References
2 Methods and Equipment for Signal Acquisition and Analysis for the Detection of Prohibited Substances
	2.1 Signal Detection and Estimation Theory
	2.2 Techniques for Detecting Prohibited Substances
		2.2.1 Trace Detection Techniques
		2.2.2 Bulk Detection Techniques
	2.3 Conclusions
	References
3 Nuclear Quadrupole Resonance Spectroscopy
	3.1 The Principle of Nuclear Quadrupole Resonance
	3.2 Signal Acquisition Methods Applied in Nuclear Quadrupole Resonance Spectroscopy
	3.3 Equipment Used in Nuclear Quadrupole Resonance Spectroscopy
	3.4 Conclusions
	Appendix
		NQR Data for Different Substances
	References
4 Signal Processing and Analysis Techniques Applied in Nuclear Quadrupole Resonance
	4.1 Classification of the Signal Processing Techniques
	4.2 Pre-Processing Techniques
		4.2.1 Signal Averaging
		4.2.2 Quadrature Detection
	4.3 Post-Processing and Analysis Techniques
		4.3.1 Single Channel Detection Techniques
		4.3.2 Multi-channel Detection Techniques
	4.4 Future Research Directions
	4.5 Conclusions
	Appendix
		Signal Processing and Analysis Techniques Classification
		Signal Post-Processing and Analysis Techniques Development Timeline
		Development of the Post-Processing Techniques
		Distribution of the Post-Processing Techniques According to the Type of Detection [1]
	References
5 Modeling of Signals Used in Nuclear Quadrupole Resonance Spectroscopy
	5.1 Free Induction Decay Models
	5.2 Echo Train Models
	5.3 Proposal of a New Echo Model
	5.4 Noise Models
	5.5 Conclusions
	References
6 Study of the NQR Signal Processing and Analysis Algorithms
	6.1 Description of the Algorithms Chosen for the Analysis
	6.2 Proposal of a Multi-criteria Detection Algorithm
	6.3 Comparative Analysis of the Signal Processing and Signal Analysis Algorithms
		6.3.1 Analysis Procedure and Performance Indicators
		6.3.2 Implementation of Comparative Analysis
		6.3.3 Comparative Analysis Results
	6.4 Conclusions
	Appendix
	References
7 Analysis of Nuclear Quadrupole Resonance Response Signals
	7.1 Description of the Analyzed Signals
	7.2 Defining the Signal Features
	7.3 Statistical Analysis of Data
	7.4 Conclusions
	Appendix
		Signal Acquisitions in Different Scanning Scenarios
		Statistical Analysis of the Data Set
	References
8 Development of Signal Analysis Algorithms for NQR Detection
	8.1 Selection of the Investigated Algorithms
	8.2 Implementation of the Signal Analysis Algorithms
	8.3 Training and Evaluation of the Signal Analysis Algorithms
	8.4 Optimization of the Proposed Algorithm
	8.5 Conclusions
	Appendix
		Source Code of the Functions Used to Generate the Deep Learning Models
		Learning Curves for the Developed Algorithms
		AlexNet Model Optimization Results
	References
9 Solutions to Improve NQR Detection
	9.1 Generalization of the Optimized Algorithm Using Transfer Learning
	9.2 Noise Reduction Using Autoencoders
	9.3 Conclusions
	Appendix
		Source Code of the Functions Used to Generate the Autoencoder Models
	References
10 Implementation of a Signal Pre-processing, Processing and Analysis System for Nuclear Quadrupole Resonance
	10.1 Hardware Implementation of the Detection System
	10.2 Software Implementation of the Detection System
		10.2.1 Embedded System Firmware
		10.2.2 Embedded System Interface
		10.2.3 Spincore Module Interface
		10.2.4 Database Interface
		10.2.5 Interface with External Applications
		10.2.6 Application Core
		10.2.7 Graphical User Interface
		10.2.8 Implementation of the Detection Application
	10.3 Evaluation of the Detection System
		10.3.1 Evaluation of the Radio Frequency Switch
		10.3.2 Detection Limits
		10.3.3 Ferrite Core Improvement
		10.3.4 Evaluation of the Signal-to-Noise Ratio
	10.4 Conclusions
	References