[Paper Review] A Review of 1D Convolutional Neural Networks toward Unknown Substance Identification in Portable Raman Spectrometer

M. Hamed Mozaffari, Li‐Lin Tay|arXiv (Cornell University)|Jun 18, 2020

Spectroscopy Techniques in Biomedical and Chemical Research60 references35 citations

TL;DR

This paper surveys how 1D CNNs are used for identifying unknown substances with portable Raman spectrometers, highlighting advantages over conventional spectral matching and considerations for handheld deployment.

ABSTRACT

Raman spectroscopy is a powerful analytical tool with applications ranging from quality control to cutting edge biomedical research. One particular area which has seen tremendous advances in the past decade is the development of powerful handheld Raman spectrometers. They have been adopted widely by first responders and law enforcement agencies for the field analysis of unknown substances. Field detection and identification of unknown substances with Raman spectroscopy rely heavily on the spectral matching capability of the devices on hand. Conventional spectral matching algorithms (such as correlation, dot product, etc.) have been used in identifying unknown Raman spectrum by comparing the unknown to a large reference database. This is typically achieved through brute-force summation of pixel-by-pixel differences between the reference and the unknown spectrum. Conventional algorithms have noticeable drawbacks. For example, they tend to work well with identifying pure compounds but less so for mixture compounds. For instance, limited reference spectra inaccessible databases with a large number of classes relative to the number of samples have been a setback for the widespread usage of Raman spectroscopy for field analysis applications. State-of-the-art deep learning methods (specifically convolutional neural networks CNNs), as an alternative approach, presents a number of advantages over conventional spectral comparison algorism. With optimization, they are ideal to be deployed in handheld spectrometers for field detection of unknown substances. In this study, we present a comprehensive survey in the use of one-dimensional CNNs for Raman spectrum identification. Specifically, we highlight the use of this powerful deep learning technique for handheld Raman spectrometers taking into consideration the potential limit in power consumption and computation ability of handheld systems.

Motivation & Objective

• Survey the use of 1D CNNs for Raman spectrum-based unknown substance identification in handheld devices.
• Highlight limitations of conventional spectral matching algorithms and how CNNs address them.
• Consider practical constraints of power and computation in portable spectrometers.
• Provide guidance on deploying CNN-based methods for field analysis and mixture identification.

Proposed method

• Review existing literature on 1D CNNs applied to Raman spectral data.
• Compare CNN-based approaches to traditional pixel-wise similarity measures.
• Discuss factors impacting handheld implementation such as power consumption and computational resources.
• Summarize findings and identify open challenges for unknown substance identification in field settings.

Experimental results

Research questions

• RQ1 What are the advantages of using 1D CNNs over conventional spectral matching for Raman spectra?
• RQ2 How do 1D CNNs perform with pure versus mixture substances in handheld Raman spectrometers?
• RQ3 What practical constraints (power, computation) affect deploying 1D CNNs on portable devices?
• RQ4 What are the gaps and open challenges in applying 1D CNNs to unknown substance identification in field scenarios?

Key findings

• 1D CNNs offer advantages in handling spectral variability and mixtures compared to traditional correlation-based methods.
• CNN-based approaches have potential to be deployed on handheld spectrometers with consideration of limited power and compute resources.
• The review highlights the need for robust datasets and model efficiency for field deployment.
• There are existing challenges related to database size, class coverage, and real-time inference in portable settings.
• The survey provides a consolidated view of how 1D CNNs have been adapted for Raman spectrum identification in portable devices.