DeepSpectra: An end-to-end deep learning approach for quantitative spectral analysis

Zhang, Xiaolei; Lin, Zhixian; Xu, Jinfan; Luo, Xuan; Ying, Yibin

doi:10.1016/j.aca.2019.01.002

Cited by 244 publications

(159 citation statements)

References 23 publications

Supporting

Mentioning

154

Contrasting

Unclassified

Order By: Relevance

“…For model 1, two fully connected layers out of three from the original LeNet architecture are kept. The adaptation of 2D-CNN architecture to 1D input was described previously, for example in Inception modules [37] according to data specificities. Using this approach, we expect to determine what model depth and hyper-parameters are optimal for MS spectra classification.…”

Section: Protocol For Evaluating Prominent 2d-cnn Architectures Adaptmentioning

confidence: 99%

“…Most of MS classification by CNNs focused on MS 2D imaging analysis [33][34][35]. Only few studies of input signal classification or regression using 1D-CNNs with vibrational spectroscopy data [36], Near-Infrared (NIR) spectroscopy data [37][38][39] or Raman spectroscopy data [26] have been published. We have found no description of their use or of transfer learning or representation learning in conjunction with 1D-MS data.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Towards CNN Representations for Small Mass Spectrometry Data Classification: From Transfer Learning to Cumulative Learning

Seddiki

Saudemont

Precioso

et al. 2020

Preprint

View full text Add to dashboard Cite

Rapid and accurate clinical diagnosis of pathological conditions remains highly challenging. A very important component of diagnosis tool development is the design of effective classification models with Mass spectrometry (MS) data. Some popular Machine Learning (ML) approaches have been investigated for this purpose but these ML models require time-consuming preprocessing steps such as baseline correction, denoising, and spectrum alignment to remove non-sample-related data artifacts. They also depend on the tedious extraction of handcrafted features, making them unsuitable for rapid analysis. Convolutional Neural Networks (CNNs) have been found to perform well under such circumstances since they can learn efficient representations from raw data without the need for costly preprocessing. However, their effectiveness drastically decreases when the number of available training samples is small, which is a common situation in medical applications. Transfer learning strategies extend an accurate representation model learnt usually on a large dataset containing many categories, to a smaller dataset with far fewer categories. In this study, we first investigate transfer learning on a 1D-CNN we have designed to classify MS data, then we develop a new representation learning method when transfer learning is not powerful enough, as in cases of low-resolution or data heterogeneity. What we propose is to train the same model through several classification tasks over various small datasets in order to accumulate generic knowledge of what MS data are, in the resulting representation. By using rat brain data as the initial training dataset, a representation learning approach can have a classification accuracy exceeding 98% for canine sarcoma cancer cells, human ovarian cancer serums, and pathogenic microorganism biotypes in 1D clinical datasets. We show for the first time the use of cumulative representation learning using datasets generated in different biological contexts, on different organisms, in different mass ranges, with different MS ionization sources, and acquired by different instruments at different resolutions. Our approach thus proposes a promising strategy for improving MS data classification accuracy when only small numbers of samples are available as a prospective cohort. The principles demonstrated in this work could even be beneficial to other domains (astronomy, archaeology...) where training samples are scarce.

show abstract

Section: Protocol For Evaluating Prominent 2d-cnn Architectures Adaptmentioning

confidence: 99%

Section: Introductionmentioning

confidence: 99%

Towards CNN Representations for Small Mass Spectrometry Data Classification: From Transfer Learning to Cumulative Learning

Seddiki

Saudemont

Precioso

et al. 2020

Preprint

View full text Add to dashboard Cite

show abstract

“…Deep learning allows models composed of multiple processing layers to learn multiple levels of representations and can discover intricate structures from high-dimensional data [19]. Recently, CNNs have been successfully used for classification tasks in IR [15], NIR [15], Raman [15,21] spectral analysis, and used for regression tasks in IR [22], NIR [22][23][24][25][26], Raman [27] spectral analysis. Recently, CNNs have been successfully used for classification tasks in IR [15], NIR [15], Raman [15,21] spectral analysis, and used for regression tasks in IR [22], NIR [22][23][24][25][26], Raman [27] spectral analysis.…”

Section: Introductionmentioning

confidence: 99%

“…Convolutional neural networks (CNNs), one of the most popular deep learning methods, have brought breakthroughs in processing images, video, speech, and audio [19], and are still growing at a rapid pace [20]. These studies have indicated that the CNN modeling method can have better model performance in some cases compared with some traditional machine learning methods [23,24,26], even when raw spectral data without any preprocessing were used as model inputs in the studies conducted by Acquarelli et al [15] and Zhang et al [24]. These studies have indicated that the CNN modeling method can have better model performance in some cases compared with some traditional machine learning methods [23,24,26], even when raw spectral data without any preprocessing were used as model inputs in the studies conducted by Acquarelli et al [15] and Zhang et al [24].…”

Section: Introductionmentioning

confidence: 99%

“…Recently, CNNs have been successfully used for classification tasks in IR [15], NIR [15], Raman [15,21] spectral analysis, and used for regression tasks in IR [22], NIR [22][23][24][25][26], Raman [27] spectral analysis. These studies have indicated that the CNN modeling method can have better model performance in some cases compared with some traditional machine learning methods [23,24,26], even when raw spectral data without any preprocessing were used as model inputs in the studies conducted by Acquarelli et al [15] and Zhang et al [24]. Moreover, Cui et al [23] have reported the "automatic preprocessing" function of the convolutional layer by visualizing convolutional layer outputs, which can give an explanation why CNN models using raw spectral data as inputs still have satisfactory performance in some reported cases.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Combining convolutional neural networks and in‐line near‐infrared spectroscopy for real‐time monitoring of the chromatographic elution process in commercial production of notoginseng total saponins

Yan

Zhang

et al. 2019

J of Separation Science

View full text Add to dashboard Cite

The chromatographic elution process is a key step in the production of notoginseng total saponins. Due to quality variability of loading samples and resin capacity decreasing over cycle time, saponins, especially the five main saponins of notoginseng total saponins, need to be monitored in real time during the elution process. In this study, convolutional neural networks, one of the most popular deep learning methods, were used to develop quantitative calibration models based on in-line near-infrared spectroscopy for notoginsenoside R 1 , ginsenosides Rg 1 , Re, Rb 1 and Rd, and their sum concentration, with root mean square error of prediction values of 0.87, 2.76, 0.60, 1.57, 0.28, and 4.99 mg/mL, respectively. Partial least squares calibration models were also developed for model performance comparison. Results show predicted concentration profiles outputted by both the convolutional neural network models and partial least squares models show agreements with the real trends defined by reference measurements, and can be used for elution process monitoring and endpoint determination. To the best of our knowledge, this is the first reported case study of combining convolutional neural networks and in-line near-infrared spectroscopy for monitoring of the chromatographic elution process in commercial production of botanical drug products. K E Y W O R D Sbotanical drugs, chromatographic elution process, convolutional neural network, deep learning, nearinfrared spectroscopy Article Related Abbreviations: BN, batch normalization; CNN, convolutional neural network; FC, fully connected; NIR, near-infrared; NTS, notoginseng total saponins; PLS, partial least squares; ReLU, rectified linear unit; RMSEP, root mean square error of prediction.[1-3] and a dietary supplement in USA [4]. Notoginseng total saponins (NTS) are the main active ingredients of Sanqi and have pharmacological activities such as anti-atherosclerotic, haemostatic, neuroprotective, anti-inflammatory, antiaging, and antioxidant effects [1,2]. Notoginsenoside R 1 , ginsenosides Rg 1 , Re, Rb 1 , and Rd are the five main saponins of NTS. These five saponins are chosen as index components for quality evaluation of NTS [5] and have been taken as qualitymarkers (Q-markers) of NTS [6].As a key step in the production of NTS, the column chromatographic process is used for removing impurities and

show abstract

NIR spectroscopy—CNN‐enabled chemometrics for multianalyte monitoring in microbial fermentation

Banerjee,

Mandal,

Jesubalan

et al. 2024

Biotech & Bioengineering

View full text Add to dashboard Cite

As the biopharmaceutical industry looks to implement Industry 4.0, the need for rapid and robust analytical characterization of analytes has become a pressing priority. Spectroscopic tools, like near‐infrared (NIR) spectroscopy, are finding increasing use for real‐time quantitative analysis. Yet detection of multiple low‐concentration analytes in microbial and mammalian cell cultures remains an ongoing challenge, requiring the selection of carefully calibrated, resilient chemometrics for each analyte. The convolutional neural network (CNN) is a puissant tool for processing complex data and making it a potential approach for automatic multivariate spectral processing. This work proposes an inception module‐based two‐dimensional (2D) CNN approach (I‐CNN) for calibrating multiple analytes using NIR spectral data. The I‐CNN model, coupled with orthogonal partial least squares (PLS) preprocessing, converts the NIR spectral data into a 2D data matrix, after which the critical features are extracted, leading to model development for multiple analytes. Escherichia coli fermentation broth was taken as a case study, where calibration models were developed for 23 analytes, including 20 amino acids, glucose, lactose, and acetate. The I‐CNN model result statistics depicted an average R2 values of prediction 0.90, external validation data set 0.86 and significantly lower root mean square error of prediction values ∼0.52 compared to conventional regression models like PLS. Preprocessing steps were applied to I‐CNN models to evaluate any augmentation in prediction performance. Finally, the model reliability was assessed via real‐time process monitoring and comparison with offline analytics. The proposed I‐CNN method is systematic and novel in extracting distinctive spectral features from a multianalyte bioprocess data set and could be adapted to other complex cell culture systems requiring rapid quantification using spectroscopy.

show abstract

DeepSpectra: An end-to-end deep learning approach for quantitative spectral analysis

Cited by 244 publications

References 23 publications

Towards CNN Representations for Small Mass Spectrometry Data Classification: From Transfer Learning to Cumulative Learning

Towards CNN Representations for Small Mass Spectrometry Data Classification: From Transfer Learning to Cumulative Learning

Combining convolutional neural networks and in‐line near‐infrared spectroscopy for real‐time monitoring of the chromatographic elution process in commercial production of notoginseng total saponins

NIR spectroscopy—CNN‐enabled chemometrics for multianalyte monitoring in microbial fermentation

Contact Info

Product

Resources

About