Convolution Network with Custom Loss Function for the Denoising of Low SNR Raman Spectra

Abstract

Raman spectroscopy is a powerful diagnostic tool in biomedical science, whereby different disease groups can be classified based on subtle differences in the cell or tissue spectra. A key component in the classification of Raman spectra is the application of multi-variate statistical models. However, Raman scattering is a weak process, resulting in a trade-off between acquisition times and signal-to-noise ratios, which has limited its more widespread adoption as a clinical tool. Typically denoising is applied to the Raman spectrum from a biological sample to improve the signal-to-noise ratio before application of statistical modeling. A popular method for performing this is Savitsky–Golay filtering. Such an algorithm is difficult to tailor so that it can strike a balance between denoising and excessive smoothing of spectral peaks, the characteristics of which are critically important for classification purposes. In this paper, we demonstrate how Convolutional Neural Networks may be enhanced with a non-standard loss function in order to improve the overall signal-to-noise ratio of spectra while limiting corruption of the spectral peaks. Simulated Raman spectra and experimental data are used to train and evaluate the performance of the algorithm in terms of the signal to noise ratio and peak fidelity. The proposed method is demonstrated to effectively smooth noise while preserving spectral features in low intensity spectra which is advantageous when compared with Savitzky–Golay filtering. For low intensity spectra the proposed algorithm was shown to improve the signal to noise ratios by up to 100% in terms of both local and overall signal to noise ratios, indicating that this method would be most suitable for low light or high throughput applications.

A Physics-Informed Convolutional Neural Network with Custom Loss Functions for Porosity Prediction in Laser Metal Deposition

“…This paper is centered on the weakly explored notion of incorporating physics-informed insights into the architecture of a deep learning model via the loss function. The design of the loss function is another important component in the design of machine learning models [33,34]. Different from previous the works [23,26], this paper focuses on the capability to combine simulated physical data with experimental data in the form of physics-informed custom loss functions built into deep learning models in the LMD process.…”

State-of-the-Art Sensors Research in Ireland

“…Barton et al [17], demonstrate how convolutional neural networks may be enhanced with a nonstandard loss function in order to improve the overall signal-to-noise ratio of Raman spectra, while limiting corruption of the spectral peaks. Simulated Raman spectra and experimental data are used to train and evaluate the performance of the algorithm in terms of the signal-to-noise ratio and peak fidelity.…”

Denoising Raman spectra using fully convolutional encoder–decoder network

“…Deep learning (DL) stands as a novel and promising approach in signal denoising [8,9] . The recent studies illustrate that artificial neural network (ANN) architectures outperform the state‐of‐the‐art techniques for Raman signal denoising, such as the S‐G filter [10,11] and wavelet denoising, [11,12] in terms of noise reduction while preserving the peaks in the Raman signals.…”

“…Barton et al [10] used five stacked convolutional layers with batch normalization and rectified linear unit (ReLU) activation. Pan et al [12] mentioned that their architecture consists of seven two‐dimensional convolutional layers with ReLU activation and a pooling layer.…”

“…Fan et al [11] used auto‐encoder architecture. In these studies, Gaussian noise was added to the actual Raman signals with high exposure times [11] or computer‐generated signals [10] were used. Based on our literature research, no study worked with real noisy Raman signals.…”