Multimodal MALDI imaging mass spectrometry for improved diagnosis of melanoma.

Imaging mass spectrometry (IMS) provides promising avenues to augment histopathological investigation with rich spatio-molecular information. We have previously developed a classification model to differentiate melanoma from nevi lesions based on IMS protein data, a task that is challenging solely by histopathologic evaluation. Most IMS-focused studies collect microscopy in tandem with IMS data, but this microscopy data is generally omitted in downstream data analysis. Microscopy, nevertheless, forms the basis for traditional histopathology and thus contains invaluable morphological information. In this work, we developed a multimodal classification pipeline that uses deep learning, in the form of a pre-trained artificial neural network, to extract the meaningful morphological features from histopathological images, and combine it with the IMS data. To test whether this deep learning-based classification strategy can improve on our previous results in classification of melanocytic neoplasia, we utilized MALDI IMS data with collected serial H&E stained sections for 331 patients, and compared this multimodal classification pipeline to classifiers using either exclusively microscopy or IMS data. The multimodal pipeline achieved the best performance, with ROC-AUCs of 0.968 vs. 0.938 vs. 0.931 for the multimodal, unimodal microscopy and unimodal IMS pipelines respectively. Due to the use of a pre-trained network to perform the morphological feature extraction, this pipeline does not require any training on large amounts of microscopy data. As such, this framework can be readily applied to improve classification performance in other experimental settings where microscopy data is acquired in tandem with IMS experiments.

Toward Omics-Scale Quantitative Mass Spectrometry Imaging of Lipids in Brain Tissue Using a Multiclass Internal Standard Mixture.

Mass spectrometry imaging (MSI) has accelerated our understanding of lipid metabolism and spatial distribution in tissues and cells. However, few MSI studies have approached lipid imaging quantitatively and those that have focused on a single lipid class. We overcome this limitation by using a multiclass internal standard (IS) mixture sprayed homogeneously over the tissue surface with concentrations that reflect those of endogenous lipids. This enabled quantitative MSI (Q-MSI) of 13 lipid classes and subclasses representing almost 200 sum-composition lipid species using both MALDI (negative ion mode) and MALDI-2 (positive ion mode) and pixel-wise normalization of each lipid species in a manner analogous to that widely used in shotgun lipidomics. The Q-MSI approach covered 3 orders of magnitude in dynamic range (lipid concentrations reported in pmol/mm2) and revealed subtle changes in distribution compared to data without normalization. The robustness of the method was evaluated by repeating experiments in two laboratories using both timsTOF and Orbitrap mass spectrometers with an ∼4-fold difference in mass resolution power. There was a strong overall correlation in the Q-MSI results obtained by using the two approaches. Outliers were mostly rationalized by isobaric interferences or the higher sensitivity of one instrument for a particular lipid species. These data provide insight into how the mass resolving power can affect Q-MSI data. This approach opens up the possibility of performing large-scale Q-MSI studies across numerous lipid classes and subclasses and revealing how absolute lipid concentrations vary throughout and between biological tissues.

Diagnosis of melanoma by imaging mass spectrometry: Development and validation of a melanoma prediction model.

BACKGROUND: The definitive diagnosis of melanocytic neoplasia using solely histopathologic evaluation can be challenging. Novel techniques that objectively confirm diagnoses are needed. This study details the development and validation of a melanoma prediction model from spatially resolved multivariate protein expression profiles generated by imaging mass spectrometry (IMS). METHODS: Three board-certified dermatopathologists blindly evaluated 333 samples. Samples with triply concordant diagnoses were included in this study, divided into a training set (n = 241) and a test set (n = 92). Both the training and test sets included various representative subclasses of unambiguous nevi and melanomas. A prediction model was developed from the training set using a linear support vector machine classification model. RESULTS: We validated the prediction model on the independent test set of 92 specimens (75 classified correctly, 2 misclassified, and 15 indeterminate). IMS detects melanoma with a sensitivity of 97.6% and a specificity of 96.4% when evaluating each unique spot. IMS predicts melanoma at the sample level with a sensitivity of 97.3% and a specificity of 97.5%. Indeterminate results were excluded from sensitivity and specificity calculations. CONCLUSION: This study provides evidence that IMS-based proteomics results are highly concordant to diagnostic results obtained by careful histopathologic evaluation from a panel of expert dermatopathologists.

Spatially aware clustering of ion images in mass spectrometry imaging data using deep learning.

Computational analysis is crucial to capitalize on the wealth of spatio-molecular information generated by mass spectrometry imaging (MSI) experiments. Currently, the spatial information available in MSI data is often under-utilized, due to the challenges of in-depth spatial pattern extraction. The advent of deep learning has greatly facilitated such complex spatial analysis. In this work, we use a pre-trained neural network to extract high-level features from ion images in MSI data, and test whether this improves downstream data analysis. The resulting neural network interpretation of ion images, coined neural ion images, is used to cluster ion images based on spatial expressions. We evaluate the impact of neural ion images on two ion image clustering pipelines, namely DBSCAN clustering, combined with UMAP-based dimensionality reduction, and k-means clustering. In both pipelines, we compare regular and neural ion images from two different MSI datasets. All tested pipelines could extract underlying spatial patterns, but the neural network-based pipelines provided better assignment of ion images, with more fine-grained clusters, and greater consistency in the spatial structures assigned to individual clusters. Additionally, we introduce the relative isotope ratio metric to quantitatively evaluate clustering quality. The resulting scores show that isotopical m/z values are more often clustered together in the neural network-based pipeline, indicating improved clustering outcomes. The usefulness of neural ion images extends beyond clustering towards a generic framework to incorporate spatial information into any MSI-focused machine learning pipeline, both supervised and unsupervised.

Unsupervised machine learning for exploratory data analysis in imaging mass spectrometry.

Imaging mass spectrometry (IMS) is a rapidly advancing molecular imaging modality that can map the spatial distribution of molecules with high chemical specificity. IMS does not require prior tagging of molecular targets and is able to measure a large number of ions concurrently in a single experiment. While this makes it particularly suited for exploratory analysis, the large amount and high-dimensional nature of data generated by IMS techniques make automated computational analysis indispensable. Research into computational methods for IMS data has touched upon different aspects, including spectral preprocessing, data formats, dimensionality reduction, spatial registration, sample classification, differential analysis between IMS experiments, and data-driven fusion methods to extract patterns corroborated by both IMS and other imaging modalities. In this work, we review unsupervised machine learning methods for exploratory analysis of IMS data, with particular focus on (a) factorization, (b) clustering, and (c) manifold learning. To provide a view across the various IMS modalities, we have attempted to include examples from a range of approaches including matrix assisted laser desorption/ionization, desorption electrospray ionization, and secondary ion mass spectrometry-based IMS. This review aims to be an entry point for both (i) analytical chemists and mass spectrometry experts who want to explore computational techniques; and (ii) computer scientists and data mining specialists who want to enter the IMS field. © 2019 The Authors. Mass Spectrometry Reviews published by Wiley Periodicals, Inc. Mass SpecRev 00:1-47, 2019.

Evaluation of Distance Metrics and Spatial Autocorrelation in Uniform Manifold Approximation and Projection Applied to Mass Spectrometry Imaging Data.

In this work, uniform manifold approximation and projection (UMAP) is applied for nonlinear dimensionality reduction and visualization of mass spectrometry imaging (MSI) data. We evaluate the performance of the UMAP algorithm on MSI data sets acquired in mouse pancreas and human lymphoma samples and compare it to those of principal component analysis (PCA), t-distributed stochastic neighbor embedding (t-SNE), and the Barnes-Hut (BH) approximation of t-SNE. Furthermore, we compare different distance metrics in (BH) t-SNE and UMAP and propose the use of spatial autocorrelation as a means of comparing the resulting low-dimensional embeddings. The results indicate that UMAP is competitive with t-SNE in terms of visualization and is well-suited for the dimensionality reduction of large (>100 000 pixels) MSI data sets. With an almost fourfold decrease in runtime, it is more scalable in comparison with the current state-of-the-art: t-SNE or the Barnes-Hut approximation of t-SNE. In what seems to be the first application of UMAP to MSI data, we assess the value of applying alternative distance metrics, such as the correlation, cosine, and the Chebyshev metric, in contrast to the traditionally used Euclidean distance metric. Furthermore, we propose "histomatch" as an additional custom distance metric for the analysis of MSI data.

Connecting imaging mass spectrometry and magnetic resonance imaging-based anatomical atlases for automated anatomical interpretation and differential analysis.

Imaging mass spectrometry (IMS) is a molecular imaging technology that can measure thousands of biomolecules concurrently without prior tagging, making it particularly suitable for exploratory research. However, the data size and dimensionality often makes thorough extraction of relevant information impractical. To help guide and accelerate IMS data analysis, we recently developed a framework that integrates IMS measurements with anatomical atlases, opening up opportunities for anatomy-driven exploration of IMS data. One example is the automated anatomical interpretation of ion images, where empirically measured ion distributions are automatically decomposed into their underlying anatomical structures. While offering significant potential, IMS-atlas integration has thus far been restricted to the Allen Mouse Brain Atlas (AMBA) and mouse brain samples. Here, we expand the applicability of this framework by extending towards new animal species and a new set of anatomical atlases retrieved from the Scalable Brain Atlas (SBA). Furthermore, as many SBA atlases are based on magnetic resonance imaging (MRI) data, a new registration pipeline was developed that enables direct non-rigid IMS-to-MRI registration. These developments are demonstrated on protein-focused FTICR IMS measurements from coronal brain sections of a Parkinson's disease (PD) rat model. The measurements are integrated with an MRI-based rat brain atlas from the SBA. The new rat-focused IMS-atlas integration is used to perform automated anatomical interpretation and to find differential ions between healthy and diseased tissue. IMS-atlas integration can serve as an important accelerator in IMS data exploration, and with these new developments it can now be applied to a wider variety of animal species and modalities. This article is part of a Special Issue entitled: MALDI Imaging, edited by Dr. Corinna Henkel and Prof. Peter Hoffmann.

Automated anatomical interpretation of ion distributions in tissue: linking imaging mass spectrometry to curated atlases.

Imaging mass spectrometry (IMS) has become a prime tool for studying the distribution of biomolecules in tissue. Although IMS data sets can become very large, computational methods have made it practically feasible to search these experiments for relevant findings. However, these methods lack access to an important source of information that many human interpretations rely upon: anatomical insight. In this work, we address this need by (1) integrating a curated anatomical data source with an empirically acquired IMS data source, establishing an algorithm-accessible link between them and (2) demonstrating the potential of such an IMS-anatomical atlas link by applying it toward automated anatomical interpretation of ion distributions in tissue. The concept is demonstrated in mouse brain tissue, using the Allen Mouse Brain Atlas as the curated anatomical data source that is linked to MALDI-based IMS experiments. We first develop a method to spatially map the anatomical atlas to the IMS data sets using nonrigid registration techniques. Once a mapping is established, a second computational method, called correlation-based querying, gives an elementary demonstration of the link by delivering basic insight into relationships between ion images and anatomical structures. Finally, a third algorithm moves further beyond both registration and correlation by providing automated anatomical interpretation of ion images. This task is approached as an optimization problem that deconstructs ion distributions as combinations of known anatomical structures. We demonstrate that establishing a link between an IMS experiment and an anatomical atlas enables automated anatomical annotation, which can serve as an important accelerator both for human and machine-guided exploration of IMS experiments.

Proteomics analysis of plasma for early diagnosis of endometriosis.

OBJECTIVE: To test the hypothesis that differential surface-enhanced laser desorption/ionization time-of-flight mass spectrometry protein or peptide expression in plasma can be used in infertile women with or without pelvic pain to predict the presence of laparoscopically and histologically confirmed endometriosis, especially in the subpopulation with a normal preoperative gynecologic ultrasound examination. METHODS: Surface-enhanced laser desorption/ionization time-of-flight mass spectrometry analysis was performed on 254 plasma samples obtained from 89 women without endometriosis and 165 women with endometriosis (histologically confirmed) undergoing laparoscopies for infertility with or without pelvic pain. Data were analyzed using least squares support vector machines and were divided randomly (100 times) into a training data set (70%) and a test data set (30%). RESULTS: Minimal-to-mild endometriosis was best predicted (sensitivity 75%, 95% confidence interval [CI] 63-89; specificity 86%, 95% CI 71-94; positive predictive value 83.6%, negative predictive value 78.3%) using a model based on five peptide and protein peaks (range 4.898-14.698 m/z) in menstrual phase samples. Moderate-to-severe endometriosis was best predicted (sensitivity 98%, 95% CI 84-100; specificity 81%, 95% CI 67-92; positive predictive value 74.4%, negative predictive value 98.6%) using a model based on five other peptide and protein peaks (range 2.189-7.457 m/z) in luteal phase samples. The peak with the highest intensity (2.189 m/z) was identified as a fibrinogen ß-chain peptide. Ultrasonography-negative endometriosis was best predicted (sensitivity 88%, 95% CI 73-100; specificity 84%, 95% CI 71-96) using a model based on five peptide peaks (range 2.058-42.065 m/z) in menstrual phase samples. CONCLUSION: A noninvasive test using proteomic analysis of plasma samples obtained during the menstrual phase enabled the diagnosis of endometriosis undetectable by ultrasonography with high sensitivity and specificity. LEVEL OF EVIDENCE: II.