Application of clustering strategy for automatic segmentation of tissue regions in mass spectrometry imaging

RATIONALE

Mass spectrometry imaging (MSI) has been widely used in biomedical research fields. Each pixel in MSI consists of a mass spectrum that reflects the molecule feature of the tissue spot. Because MSI contains high-dimensional datasets, it is highly desired to develop computational methods for data mining and constructing tissue segmentation maps.

METHODS

To visualize different tissue regions based on mass spectrum features and improve the efficiency in processing enormous data, we proposed a computational strategy that consists of four procedures including preprocessing, data reduction, clustering, and quantitative validation.

RESULTS

In this study, we examined the combination of t-distributed stochastic neighbor embedding (t-SNE) and hierarchical clustering (HC) for MSI data analysis. Using publicly available MSI datasets, one dataset of mouse urinary bladder, and one dataset of human colorectal cancer, we demonstrated that the generated tissue segmentation maps from this combination were superior to other data reduction and clustering algorithms. Using the staining image as a reference, we assessed the performance of clustering algorithms with external and internal clustering validation measures, including purity, adjusted Rand index (ARI), Davies-Bouldin index (DBI), and spatial aggregation index (SAI). The result indicated that SAI delivered excellent performance for automatic segmentation of tissue regions in MSI.

CONCLUSIONS

We used a clustering algorithm to construct tissue automatic segmentation in MSI datasets. The performance was evaluated by comparing it with the stained image and calculating clustering validation indexes. The results indicated that SAI is important for automatic tissue segmentation in MSI, different from traditional clustering validation measures. Compared to the reports that used internal clustering validation measures such as DBI, our method offers more effective evaluation of clustering results for MSI segmentation. We envision that the proposed automatic image segmentation strategy can facilitate deep learning in molecular feature extraction and biomarker discovery for the biomedical applications of MSI.

Biological mass spectrometry enables spatiotemporal 'omics: From tissues to cells to organelles

Biological processes unfold across broad spatial and temporal dimensions, and measurement of the underlying molecular world is essential to their understanding. Interdisciplinary efforts advanced mass spectrometry (MS) into a tour de force for assessing virtually all levels of the molecular architecture, some in exquisite detection sensitivity and scalability in space-time. In this review, we offer vignettes of milestones in technology innovations that ushered sample collection and processing, chemical separation, ionization, and 'omics analyses to progressively finer resolutions in the realms of tissue biopsies and limited cell populations, single cells, and subcellular organelles. Also highlighted are methodologies that empowered the acquisition and analysis of multidimensional MS data sets to reveal proteomes, peptidomes, and metabolomes in ever-deepening coverage in these limited and dynamic specimens. In pursuit of richer knowledge of biological processes, we discuss efforts pioneering the integration of orthogonal approaches from molecular and functional studies, both within and beyond MS. With established and emerging community-wide efforts ensuring scientific rigor and reproducibility, spatiotemporal MS emerged as an exciting and powerful resource to study biological systems in space-time.

Next-generation pathology practices with mass spectrometry imaging

Mass spectrometry imaging (MSI) is a powerful technique that reveals the spatial distribution of various molecules in biological samples, and it is widely used in pathology-related research. In this review, we summarize common MSI techniques, including matrix-assisted laser desorption/ionization and desorption electrospray ionization MSI, and their applications in pathological research, including disease diagnosis, microbiology, and drug discovery. We also describe the improvements of MSI, focusing on the accumulation of imaging data sets, expansion of chemical coverage, and identification of biological significant molecules, that have prompted the evolution of MSI to meet the requirements of pathology practices. Overall, this review details the applications and improvements of MSI techniques, demonstrating the potential of integrating MSI techniques into next-generation pathology practices.

DiviK: divisive intelligent K-means for hands-free unsupervised clustering in big biological data

BACKGROUND

Investigating molecular heterogeneity provides insights into tumour origin and metabolomics. The increasing amount of data gathered makes manual analyses infeasible-therefore, automated unsupervised learning approaches are utilised for discovering tissue heterogeneity. However, automated analyses require experience setting the algorithms' hyperparameters and expert knowledge about the analysed biological processes. Moreover, feature engineering is needed to obtain valuable results because of the numerous features measured.

RESULTS

We propose DiviK: a scalable stepwise algorithm with local data-driven feature space adaptation for segmenting high-dimensional datasets. The algorithm is compared to the optional solutions (regular k-means, spatial and spectral approaches) combined with different feature engineering techniques (None, PCA, EXIMS, UMAP, Neural Ions). Three quality indices: Dice Index, Rand Index and EXIMS score, focusing on the overall composition of the clustering, coverage of the tumour region and spatial cluster consistency, are used to assess the quality of unsupervised analyses. Algorithms were validated on mass spectrometry imaging (MSI) datasets-2D human cancer tissue samples and 3D mouse kidney images. DiviK algorithm performed the best among the four clustering algorithms compared (overall quality score 1.24, 0.58 and 162 for d(0, 0, 0), d(1, 1, 1) and the sum of ranks, respectively), with spectral clustering being mostly second. Feature engineering techniques impact the overall clustering results less than the algorithms themselves (partial [Formula: see text] effect size: 0.141 versus 0.345, Kendall's concordance index: 0.424 versus 0.138 for d(0, 0, 0)).

CONCLUSIONS

DiviK could be the default choice in the exploration of MSI data. Thanks to its unique, GMM-based local optimisation of the feature space and deglomerative schema, DiviK results do not strongly depend on the feature engineering technique applied and can reveal the hidden structure in a tissue sample. Additionally, DiviK shows high scalability, and it can process at once the big omics data with more than 1.5 mln instances and a few thousand features. Finally, due to its simplicity, DiviK is easily generalisable to an even more flexible framework. Therefore, it is helpful for other -omics data (as single cell spatial transcriptomic) or tabular data in general (including medical images after appropriate embedding). A generic implementation is freely available under Apache 2.0 license at https://github.com/gmrukwa/divik .

Quantifying Spatial Heterogeneity of Tumor-Infiltrating Lymphocytes to Predict Survival of Individual Cancer Patients

Tumor-infiltrating lymphocytes (TILs), identified on HE-stained histopathological images in the cancer area, are indicators of the adaptive immune response against cancers and play a major role in personalized cancer immunotherapy. Recent works indicate that the spatial organization of TILs may be prognostic of disease-specific survival and recurrence. However, there are a limited number of methods that were proposed and tested in analyses of the spatial structure of TILs. In this work, we evaluated 14 different spatial measures, including the one developed for other omics data, on 10,532 TIL maps from 23 cancer types in terms of reproducibility, uniqueness, and impact on patient survival. For each spatial measure, 16 different scenarios for the definition of prognostic factor were tested. We found no difference in survival prediction when TIL maps were stored as binary images or continuous TIL probability scores. When spatial measures were discretized into a low and high category, a higher correlation with survival was observed. Three measures with the highest cancer prognosis capability were spatial autocorrelation, GLCM M1, and closeness centrality. Most of the tested measures could be further tuned to increase prediction performance.

Deterministic Tractography Analysis of Rat Brain Using SIGMA Atlas in 9.4T MRI

Preclinical studies using rodents have been the choice for many neuroscience researchers due totheir close reflection of human biology. In particular, research involving rodents has utilized MRI to accurately identify brain regions and characteristics by acquiring high resolution cavity images with different contrasts non-invasively, and this has resulted in high reproducibility and throughput. In addition, tractographic analysis using diffusion tensor imaging to obtain information on the neural structure of white matter has emerged as a major methodology in the field of neuroscience due to its contribution in discovering significant correlations between altered neural connections and various neurological and psychiatric diseases. However, unlike image analysis studies with human subjects where a myriad of human image analysis programs and procedures have been thoroughly developed and validated, methods for analyzing rat image data using MRI in preclinical research settings have seen significantly less developed. Therefore, in this study, we present a deterministic tractographic analysis pipeline using the SIGMA atlas for a detailed structural segmentation and structural connectivity analysis of the rat brain’s structural connectivity. In addition, the structural connectivity analysis pipeline presented in this study was preliminarily tested on normal and stroke rat models for initial observation.

A versatile deep learning architecture for classification and label-free prediction of hyperspectral images

Hyperspectral imaging is a technique that provides rich chemical or compositional information not regularly available to traditional imaging modalities such as intensity imaging or color imaging based on the reflection, transmission, or emission of light. Analysis of hyperspectral imaging often relies on machine learning methods to extract information. Here, we present a new flexible architecture, the U-within-U-Net, that can perform classification, segmentation, and prediction of orthogonal imaging modalities on a variety of hyperspectral imaging techniques. Specifically, we demonstrate feature segmentation and classification on the Indian Pines hyperspectral dataset and simultaneous location prediction of multiple drugs in mass spectrometry imaging of rat liver tissue. We further demonstrate label-free fluorescence image prediction from hyperspectral stimulated Raman scattering microscopy images. The applicability of the U-within-U-Net architecture on diverse datasets with widely varying input and output dimensions and data sources suggest that it has great potential in advancing the use of hyperspectral imaging across many different application areas ranging from remote sensing, to medical imaging, to microscopy.

Interactive Visual Analysis of Mass Spectrometry Imaging Data Using Linear and Non-Linear Embeddings

Mass spectrometry imaging (MSI) is an imaging technique used in analytical chemistry to study the molecular distribution of various compounds at a micro-scale level. For each pixel, MSI stores a mass spectrum obtained by measuring signal intensities of thousands of mass-to-charge ratios (m/z-ratios), each linked to an individual molecular ion species. Traditional analysis tools focus on few individual m/z-ratios, which neglects most of the data. Recently, clustering methods of the spectral information have emerged, but faithful detection of all relevant image regions is not always possible. We propose an interactive visual analysis approach that considers all available information in coordinated views of image and spectral space visualizations, where the spectral space is treated as a multi-dimensional space. We use non-linear embeddings of the spectral information to interactively define clusters and respective image regions. Of particular interest is, then, which of the molecular ion species cause the formation of the clusters. We propose to use linear embeddings of the clustered data, as they allow for relating the projected views to the given dimensions. We document the effectiveness of our approach in analyzing matrix-assisted laser desorption/ionization (MALDI-2) imaging data with ground truth obtained from histological images.

Spatial Metabolomics and Imaging Mass Spectrometry in the Age of Artificial Intelligence

Spatial metabolomics is an emerging field of omics research that has enabled localizing metabolites, lipids, and drugs in tissue sections, a feat considered impossible just two decades ago. Spatial metabolomics and its enabling technology-imaging mass spectrometry-generate big hyper-spectral imaging data that have motivated the development of tailored computational methods at the intersection of computational metabolomics and image analysis. Experimental and computational developments have recently opened doors to applications of spatial metabolomics in life sciences and biomedicine. At the same time, these advances have coincided with a rapid evolution in machine learning, deep learning, and artificial intelligence, which are transforming our everyday life and promise to revolutionize biology and healthcare. Here, we introduce spatial metabolomics through the eyes of a computational scientist, review the outstanding challenges, provide a look into the future, and discuss opportunities granted by the ongoing convergence of human and artificial intelligence.

LipostarMSI: Comprehensive, Vendor-Neutral Software for Visualization, Data Analysis, and Automated Molecular Identification in Mass Spectrometry Imaging

Mass Spectrometry Imaging (MSI) is an established and powerful MS technique that enables molecular mapping of tissues and cells finding widespread applications in academic, medical, and pharmaceutical industries. As both the applications and MSI technology have undergone rapid growth and improvement, the challenges associated both with analyzing large datasets and identifying the many detected molecular species have become apparent. The lack of readily available and comprehensive software covering all necessary data analysis steps has further compounded this challenge. To address this issue we developed LipostarMSI, comprehensive and vendor-neutral software for targeted and untargeted MSI data analysis. Through user-friendly implementation of image visualization and co-registration, univariate and multivariate image and spectral analysis, and for the first time, advanced lipid, metabolite, and drug metabolite (MetID) automated identification, LipostarMSI effectively streamlines biochemical interpretation of the data. Here, we introduce LipostarMSI and case studies demonstrating the versatility and many capabilities of the software.

Chemometric Strategies for Sensitive Annotation and Validation of Anatomical Regions of Interest in Complex Imaging Mass Spectrometry Data

Imaging mass spectrometry (IMS) is a promising new chemical imaging modality that generates a large body of complex imaging data, which in turn can be approached using multivariate analysis approaches for image analysis and segmentation. Processing IMS raw data is critically important for proper data interpretation and has significant effects on the outcome of data analysis, in particular statistical modeling. Commonly, data processing methods are chosen based on rational motivations rather than comparative metrics, though no quantitative measures to assess and compare processing options have been suggested. We here present a data processing and analysis pipeline for IMS data interrogation, processing and ROI annotation, segmentation, and validation. This workflow includes (1) objective evaluation of processing methods for IMS datasets based on multivariate analysis using PCA. This was then followed by (2) ROI annotation and classification through region-based active contours (AC) segmentation based on the PCA component scores matrix. This provided class information for subsequent (3) OPLS-DA modeling to evaluate IMS data processing based on the quality metrics of their respective multivariate models and for robust quantification of ROI-specific signal localization. This workflow provides an unbiased strategy for sensitive annotation of anatomical regions of interest combined with quantitative comparison of processing procedures for multivariate analysis allowing robust ROI annotation and quantification of the associated molecular histology.

Clusterwise Peak Detection and Filtering Based on Spatial Distribution To Efficiently Mine Mass Spectrometry Imaging Data

Mass spectrometry imaging (MSI) has the potential to reveal the localization of thousands of biomolecules such as metabolites and lipids in tissue sections. The increase in both mass and spatial resolution of today’s instruments brings on considerable challenges in terms of data processing; accurately extracting meaningful signals from the large data sets generated by MSI without losing information that could be clinically relevant is one of the most fundamental tasks of analysis software. Ion images of the biomolecules are generated by visualizing their intensities in 2-D space using mass spectra collected across the tissue section. The intensities are often calculated by summing each compound’s signal between predefined sets of borders (bins) in the m/z dimension. This approach, however, can result in mixed signals from different compounds in the same bin or splitting the signal from one compound between two adjacent bins, leading to low quality ion images. To remedy this problem, we propose a novel data processing approach. Our approach consists of a sensitive peak detection method able to discover both faint and localized signals by utilizing clusterwise kernel density estimates (KDEs) of peak distributions. We show that our method can recall more ground-truth molecules, molecule fragments, and isotopes than existing methods based on binning. Furthermore, it automatically detects previously reported molecular ions of lipids, including those close in m/z, in an experimental data set.

Recent Technological Advances in the Mass Spectrometry-based Nanomedicine Studies: An Insight from Nanoproteomics

Nanoscience becomes one of the most cutting-edge research directions in recent years since it is gradually matured from basic to applied science. Nanoparticles (NPs) and nanomaterials (NMs) play important roles in various aspects of biomedicine science, and their influences on the environment have caused a whole range of uncertainties which require extensive attention. Due to the quantitative and dynamic information provided for human proteome, mass spectrometry (MS)-based quantitative proteomic technique has been a powerful tool for nanomedicine study. In this article, recent trends of progress and development in the nanomedicine of proteomics were discussed from quantification techniques and publicly available resources or tools. First, a variety of popular protein quantification techniques including labeling and label-free strategies applied to nanomedicine studies are overviewed and systematically discussed. Then, numerous protein profiling tools for data processing and postbiological statistical analysis and publicly available data repositories for providing enrichment MS raw data information sources are also discussed.

High-resolution atmospheric-pressure MALDI mass spectrometry imaging workflow for lipidomic analysis of late fetal mouse lungs

Mass spectrometry imaging (MSI) provides label-free, non-targeted molecular and spatial information of the biomolecules within tissue. Lipids play important roles in lung biology, e.g. as surfactant, preventing alveolar collapse during normal and forced respiration. Lipidomic characterization of late fetal mouse lungs at day 19 of gestation (E19) has not been performed yet. In this study we employed high-resolution atmospheric pressure scanning microprobe matrix-assisted laser desorption/ionization MSI for the lipidomic analysis of E19 mouse lungs. Molecular species of different lipid classes were imaged in E19 lung sections at high spatial and mass resolution in positive- and negative-ion mode. Lipid species were characterized based on accurate mass and on-tissue tandem mass spectrometry. In addition, a dedicated sample preparation protocol, homogenous deposition of matrices on tissue surfaces and data processing parameters were optimized for the comparison of signal intensities of lipids between different tissue sections of E19 lungs of wild type and Pex11β-knockout mice. Our study provides lipid information of E19 mouse lungs, optimized experimental and data processing strategies for the direct comparison of signal intensities of metabolites (lipids) among the tissue sections from MSI experiments. To best of our knowledge, this is the first MSI and lipidomic study of E19 mouse lungs.

Drug-Homogeneity Index in Mass-Spectrometry Imaging

Enhancing drug penetration in solid tumors is an interesting clinical issue of considerable importance. In preclinical research, mass-spectrometry imaging is a promising technique for visualizing drug distribution in tumors under different treatment conditions and its application in this field is rapidly increasing. However, in view of the huge variability among MSI data sets, drug homogeneity is usually manually assessed by an expert, and this approach is biased by interobserver variability and lacks reproducibility. We propose a new texture-based feature, the drug-homogeneity index (DHI), which provides an objective, automated measure of drug homogeneity in MSI data. A simulation study on synthetic data sets showed that previously known texture features do not give an accurate picture of intratumor drug-distribution patterns and are easily influenced by the tumor-tissue morphology. The DHI has been used to study the distribution profile of the anticancer drug paclitaxel in various xenograft models, which were either pretreated or not pretreated with antiangiogenesis compounds. The conclusion is that drug homogeneity is better in the pretreated condition, which is in agreement with previous experimental findings published by our group. This study shows that DHI could be useful in preclinical studies as a new parameter for the evaluation of protocols for better drug penetration.

From chromatogram to analyte to metabolite. How to pick horses for courses from the massive web resources for mass spectral plant metabolomics

The grand challenge currently facing metabolomics is the expansion of the coverage of the metabolome from a minor percentage of the metabolic complement of the cell toward the level of coverage afforded by other post-genomic technologies such as transcriptomics and proteomics. In plants, this problem is exacerbated by the sheer diversity of chemicals that constitute the metabolome, with the number of metabolites in the plant kingdom generally considered to be in excess of 200 000. In this review, we focus on web resources that can be exploited in order to improve analyte and ultimately metabolite identification and quantification. There is a wide range of available software that not only aids in this but also in the related area of peak alignment; however, for the uninitiated, choosing which program to use is a daunting task. For this reason, we provide an overview of the pros and cons of the software as well as comments regarding the level of programing skills required to effectively exploit their basic functions. In addition, the torrent of available genome and transcriptome sequences that followed the advent of next-generation sequencing has opened up further valuable resources for metabolite identification. All things considered, we posit that only via a continued communal sharing of information such as that deposited in the databases described within the article are we likely to be able to make significant headway toward improving our coverage of the plant metabolome.

FDR-controlled metabolite annotation for high-resolution imaging mass spectrometry

High-mass-resolution imaging mass spectrometry promises to localize hundreds of metabolites in tissues, cell cultures, and agar plates with cellular resolution, but it is hampered by the lack of bioinformatics tools for automated metabolite identification. We report pySM, a framework for false discovery rate (FDR)-controlled metabolite annotation at the level of the molecular sum formula, for high-mass-resolution imaging mass spectrometry (https://github.com/alexandrovteam/pySM). We introduce a metabolite-signal match score and a target-decoy FDR estimate for spatial metabolomics.

A year (2014-2015) of plants in Proteomics journal. Progress in wet and dry methodologies, moving from protein catalogs, and the view of classic plant biochemists

The present review is an update of the previous one published in Proteomics 2015 Reviews special issue [Jorrin-Novo, J. V. et al., Proteomics 2015, 15, 1089-1112] covering the July 2014-2015 period. It has been written on the bases of the publications that appeared in Proteomics journal during that period and the most relevant ones that have been published in other high-impact journals. Methodological advances and the contribution of the field to the knowledge of plant biology processes and its translation to agroforestry and environmental sectors will be discussed. This review has been organized in four blocks, with a starting general introduction (literature survey) followed by sections focusing on the methodology (in vitro, in vivo, wet, and dry), proteomics integration with other approaches (systems biology and proteogenomics), biological information, and knowledge (cell communication, receptors, and signaling), ending with a brief mention of some other biological and translational topics to which proteomics has made some contribution.

Probabilistic Segmentation of Mass Spectrometry (MS) Images Helps Select Important Ions and Characterize Confidence in the Resulting Segments

Mass spectrometry imaging is a powerful tool for investigating the spatial distribution of chemical compounds in a biological sample such as tissue. Two common goals of these experiments are unsupervised segmentation of images into newly discovered homogeneous segments and supervised classification of images into predefined classes. In both cases, the important secondary goals are to characterize the uncertainty associated with the segmentation and with the classification and to characterize the spectral features that define each segment or class. Recent analysis methods have focused on the spatial structure of the data to improve results. However, they either do not address these secondary goals or do this with separate post hoc procedures.

We introduce spatial shrunken centroids, a statistical model-based framework for both supervised classification and unsupervised segmentation. It takes as input sets of previously detected, aligned, quantified, and normalized spectral features and expresses both spatial and multivariate nature of the data using probabilistic modeling. It selects informative subsets of spectral features that define each unsupervised segment or supervised class and quantifies and visualizes the uncertainty in spatial segmentations and in tissue classification. In the unsupervised setting, it also guides the choice of an appropriate number of segments. We demonstrate the usefulness of this framework in a supervised human renal cell carcinoma experimental dataset and several unsupervised experimental datasets, including a pig fetus cross-section, three rodent brains, and a controlled image with known ground truth. This framework is available for use within the open-source R package Cardinal as part of a full pipeline for the processing, visualization, and statistical analysis of mass spectrometry imaging experiments.

A new peak detection algorithm for MALDI mass spectrometry data based on a modified Asymmetric Pseudo-Voigt model

BACKGROUND

Mass Spectrometry (MS) is a ubiquitous analytical tool in biological research and is used to measure the mass-to-charge ratio of bio-molecules. Peak detection is the essential first step in MS data analysis. Precise estimation of peak parameters such as peak summit location and peak area are critical to identify underlying bio-molecules and to estimate their abundances accurately. We propose a new method to detect and quantify peaks in mass spectra. It uses dual-tree complex wavelet transformation along with Stein's unbiased risk estimator for spectra smoothing. Then, a new method, based on the modified Asymmetric Pseudo-Voigt (mAPV) model and hierarchical particle swarm optimization, is used for peak parameter estimation.

RESULTS

Using simulated data, we demonstrated the benefit of using the mAPV model over Gaussian, Lorentz and Bi-Gaussian functions for MS peak modelling. The proposed mAPV model achieved the best fitting accuracy for asymmetric peaks, with lower percentage errors in peak summit location estimation, which were 0.17% to 4.46% less than that of the other models. It also outperformed the other models in peak area estimation, delivering lower percentage errors, which were about 0.7% less than its closest competitor - the Bi-Gaussian model. In addition, using data generated from a MALDI-TOF computer model, we showed that the proposed overall algorithm outperformed the existing methods mainly in terms of sensitivity. It achieved a sensitivity of 85%, compared to 77% and 71% of the two benchmark algorithms, continuous wavelet transformation based method and Cromwell respectively.

CONCLUSIONS

The proposed algorithm is particularly useful for peak detection and parameter estimation in MS data with overlapping peak distributions and asymmetric peaks. The algorithm is implemented using MATLAB and the source code is freely available at http://mapv.sourceforge.net.