R.A.R. Moens | TU Delft Repository

Structured and Low-Rank Decompositions for Large-Scale Imaging Datasets

Doctoral thesis (2026) - R.A.R. Moens, R. Van de Plas, B. De Schutter

A common strategy to address new scientific challenges consists of abstracting the underlying problem, recasting it to an existing problem formulation and applying an established methodology. In this dissertation, we offer a variation on this familiar academic theme. The setting we will focus on is primarily found within image-acquiring instruments, characterized by producing vast quantities of data, from several hundreds up to more than half a million images per experiment. The challenges that we address throughout this work will mainly consist of (a) reducing dimensionality and (b) denoising, which have a direct and significant impact on the analysis and thus interpretation of these extensive image sets.
We investigate computational methods for two specific imaging instruments: (1) a time-of-flight imaging mass spectrometer, employed in biochemical research to visualize molecular distributions across very small organic tissues, and (2) a mid-infrared imager, utilized in astronomical research to study very large protostars, temperate exoplanets, and objects within our solar system. Despite their considerable promise in acquiring detailed molecular maps and critical astronomical insights, respectively, the practical analysis and interpretation of their image sets face substantial obstacles, namely their dimensionality and the effect of noise. Addressing these challenges may involve drawing on existing computational and storage capacity and harnessing any available prior information or problem-specific structure. Fortunately, analytical solutions to the obstacles across imaging instruments often bear a resemblance to each other, as we distil them to abstract mathematical models and eventually formulate those problems as optimization problems.
The computational methods we are interested in are so-called low-rankmethods, they can simultaneously provide insight in data structure (analysis), as well as reduce the dimensionality of the data and denoise it.... ...

A common strategy to address new scientific challenges consists of abstracting the underlying problem, recasting it to an existing problem formulation and applying an established methodology. In this dissertation, we offer a variation on this familiar academic theme. The setting we will focus on is primarily found within image-acquiring instruments, characterized by producing vast quantities of data, from several hundreds up to more than half a million images per experiment. The challenges that we address throughout this work will mainly consist of (a) reducing dimensionality and (b) denoising, which have a direct and significant impact on the analysis and thus interpretation of these extensive image sets.
We investigate computational methods for two specific imaging instruments: (1) a time-of-flight imaging mass spectrometer, employed in biochemical research to visualize molecular distributions across very small organic tissues, and (2) a mid-infrared imager, utilized in astronomical research to study very large protostars, temperate exoplanets, and objects within our solar system. Despite their considerable promise in acquiring detailed molecular maps and critical astronomical insights, respectively, the practical analysis and interpretation of their image sets face substantial obstacles, namely their dimensionality and the effect of noise. Addressing these challenges may involve drawing on existing computational and storage capacity and harnessing any available prior information or problem-specific structure. Fortunately, analytical solutions to the obstacles across imaging instruments often bear a resemblance to each other, as we distil them to abstract mathematical models and eventually formulate those problems as optimization problems.
The computational methods we are interested in are so-called low-rankmethods, they can simultaneously provide insight in data structure (analysis), as well as reduce the dimensionality of the data and denoise it....

Preserving full spectrum information in imaging mass spectrometry data reduction

Journal article (2025) - Roger A.R. Moens, Lukasz G. Migas, Jacqueline M. Van Ardenne, Eric P. Skaar, Jeffrey M. Spraggins, Raf Van de Plas

Motivation: Imaging mass spectrometry (IMS) has become an important tool for molecular characterization of biological tissue. However, IMS experiments tend to yield large datasets, routinely recording over 200 000 ion intensity values per mass spectrum and more than 100 000 pixels, i.e. spectra, per dataset. Traditionally, IMS data size challenges have been addressed by feature selection or extraction, such as by peak picking and peak integration. Selective data reduction techniques such as peak picking only retain certain parts of a mass spectrum, and often these describe only medium-to-high-abundance species. Since lower-intensity peaks and, for example, near-isobar species are sometimes missed, selective methods can potentially bias downstream analysis toward a subset of species in the data rather than considering all species measured. Results: We present an alternative to selective data reduction of IMS data that achieves similar data size reduction while better conserving the ion intensity profiles across all recorded m/z-bins, thereby preserving full spectrum information. Our method utilizes a low-rank matrix completion model combined with a randomized sparse-format-aware algorithm to approximate IMS datasets. This representation offers reduced dimensionality and a data footprint comparable to peak picking but also captures complete spectral profiles, enabling comprehensive analysis and compression. We demonstrate improved preservation of lower signal-to-noise ratio signals and near-isobars, mitigation of selection bias, and reduced information loss compared to current state-of-the-art data reduction methods in IMS. Availability and implementation: The source code is available at https://github.com/vandeplaslab/full_profile and data are available at https://doi.org/10.4121/a6efd47a-b4ec-493e-a742-70e8a369f788. ...

Advanced Dimensionality Reduction for Imaging Mass Spectrometry of Human Eye Tissue through Low-Rank Modeling with Sparse and Dense Residuals

Journal article (2025) - R.A.R. Moens, L.G. Migas, David M G Anderson, Jeffrey D. Messinger, Olga S. Ovchinnikova, Richard M. Caprioli, Christine A. Curcio, Kevin L. Schey, Jeffrey M. Spraggins, Raf Van de Plas

Imaging mass spectrometry (IMS) yields high-dimensional and large data sets commonly exceeding 100,000 pixels, each reporting a mass spectrum of 200,000 intensity values or more. Reducing the dimensionality and size of IMS data is often necessary to enable downstream analysis, and matrix-factorization-based approaches are often used for this purpose. However, the model underlying most of these techniques, decomposing measurements into the sum of a low-rank term (presumed signal) and a small entry-wise residual term (presumed noise), is often not optimal for IMS. For example, while spatially or spectrally sparse signals are common in IMS data, they can heavily distort the low-rank approximation. Therefore, we propose capturing the IMS data structure using low-rank models that, in addition to a dense residual, allow for sparse variation to be captured separately. We implement two such methods, principal component pursuit (PCP) and stable principal component pursuit (SPCP), apply them to IMS data, and compare them to a classical factorization method, principal component analysis (PCA). We investigate their dimensionality and noise reduction performance on MALDI Q-TOF IMS measurements of human cornea and retina tissue since the human eye is a complex organ with lots of small, tightly packed tissue substructures that are spatially sparse. Our results suggest that if parameters are set adequately, PCP and SPCP enable stronger dimensionality reduction and higher compression of IMS data compared to PCA, while concurrently reducing signal overestimation. ...

Correction to

Advances and prospects for the Human BioMolecular Atlas Program (HuBMAP) (Nature Cell Biology, (2023), 25, 8, (1089-1100), 10.1038/s41556-023-01194-w)

Journal article (2024) - Sanjay Jain, Liming Pei, Joana P. Gonçalves, Huiping Liu, Paul Robson, Raf Van de Plas, Roy Lardenoije, Lukasz G. Migas, Roger Moens, More authors...

Correction to: Nature Cell Biologyhttps://doi.org/10.1038/s41556-023-01194-w. Published online 19 July 2023. In the version of this article originally published, the name of Tianyang Xu was misspelled as Tiangyang Xu. The name has been corrected in the HTML and PDF versions of the article. ...

Advances and prospects for the Human BioMolecular Atlas Program (HuBMAP)

Journal article (2023) - Sanjay Jain, Liming Pei, Jeffrey M. Spraggins, Michael Angelo, Joana P. Gonçalves, Raf Van de Plas, R. Lardenoije, L.G. Migas, R.A.R. Moens, More authors...

The Human BioMolecular Atlas Program (HuBMAP) aims to create a multi-scale spatial atlas of the healthy human body at single-cell resolution by applying advanced technologies and disseminating resources to the community. As the HuBMAP moves past its first phase, creating ontologies, protocols and pipelines, this Perspective introduces the production phase: the generation of reference spatial maps of functional tissue units across many organs from diverse populations and the creation of mapping tools and infrastructure to advance biomedical research. ...