Modified PCA and PLS: Towards a better classification in Raman spectroscopy–based biological applications

Abstract

Raman spectra of biological samples often exhibit variations originating from changes of spectrometers, measurement conditions, and cultivation conditions. Such unwanted variations make a classification extremely challenging, especially if they are more significant compared with the differences between groups to be separated. A classifier is prone to such unwanted variations (ie, intragroup variations) and can fail to learn the patterns that can help separate different groups (ie, intergroup differences). This often leads to a poor generalization performance and a degraded transferability of the trained model. A natural solution is to separate the intragroup variations from the intergroup differences and build the classifier based on merely the latter information, for example, by a well‐designed feature extraction. This forms the idea of this contribution. Herein, we modified two commonly applied feature extraction approaches, principal component analysis (PCA) and partial least squares (PLS), in order to extract merely the features representing the intergroup differences. Both of the methods were verified with two Raman spectral datasets measured from bacterial cultures and colon tissues of mice, respectively. In comparison to ordinary PCA and PLS, the modified PCA was able to improve the prediction on the testing data that bears significant difference to the training data, while the modified PLS could help avoid overfitting and lead to a more stable classification.

Machine learning utilising spectral derivative data improves cellular health classification through hyperspectral infra-red spectroscopy

“…Partial least squares (PLS) is an additional dimensionality reductions methodology that considers the correlation between both the dependant and independent variables, unlike PCA which only considers the independent variables [40]. Such methods have also been further developed with modified versions for analysis of Raman spectroscopy, with improved classification shown [41]. Dimensionality reduction could also be explored further through the use of autoencoders, which also consider non-linear contributions, although these require additional computation due to its neural network design [42].…”

Surface-enhanced Raman spectroscopy for bioanalysis and diagnosis

“…intergroup differences). Support-vector machines 192 and complex data processing 193,194 are able to provide remarkable results by using dedicated software. On the other hand, the multi-level computational models (PCA-PLS-LDA or PLS-LDA-LDA) that are checked and validated by using data from an independent biological replicate produce high sensitivity and even higher specificity with respect to the prediction based on minor spectral changes.…”

A rapid food chain approach for authenticity screening: The development, validation and transferability of a chemometric model using two handheld near infrared spectroscopy (NIRS) devices

“…As PCA is an unsupervised technique, the classification does not necessarily relate to the principal component patterns. In this way, it is not prone to overfitting, which is regularly seen with PLS-DA analysis, due to the influence of projecting the data in relation to the best separation of classification groups [ 37 ]. Fig.…”