arXiv subjects

Paul Stewart

Publications and source records attributed to Paul Stewart.

11 recordsLinked to original sources

arXiv2026

Non-destructive Identification of Oyster Species is possible from Hyperspectral Images with Machine Learning

Differentiating between oyster species is important for developing new commercial oyster species suited to production systems and is critical for traceability in seafood supply chains. Common methods, such as DNA profiling, are destructive and time consuming. The possibility of using hyperspectral imaging (HSI) for discriminating between Black-Lip rock (BL) and Sydney rock (SR) oysters was investigated. Live BL and SR samples (N = 156) were scanned with a HSI camera (950-2515nm). Partial Least Square Discriminant Analysis and Convolutional Neural Networks were trained with Monte Carlo Cross Validation to distinguish BL and SR oysters from the spectral reflectance of their left and rights valves. The PLS-DA model successfully distinguished between the species from both the left and right valves with a median test set classification accuracy of 100%, out performing the CNN with 83% and 96% respectively. Elemental and mineralogical composition in the surface and cross-section of oyster valves were measured with electron microscopy. Analysis of the right valve revealed a greater number of layers in BL compared to SR (4 vs 2). The concentrations of carbon and oxygen varied in the outer layer of the right valves, with BL being rich in carbon and SR being rich in oxygen. The variation in carbon and oxygen concentrations observed between BL and SR right valves may reflect differences in the relative abundance or composition of chitin and glycoproteins. This is supported by model-derived wavelength importance corresponding to vibrational modes of functional groups characteristic of these compounds. Transmittance analysis revealed that light was transmitted through the valves, around the valve edges, indicating that the spectral signatures may have been influenced by the other valve or the meat. Ultimately, the findings highlight an effective rapid, non-destructive methodology for oyster species.

cs.LG↗

arXiv2024

Embedding-based Multimodal Learning on Pan-Squamous Cell Carcinomas for Improved Survival Outcomes

Cancer clinics capture disease data at various scales, from genetic to organ level. Current bioinformatic methods struggle to handle the heterogeneous nature of this data, especially with missing modalities. We propose PARADIGM, a Graph Neural Network (GNN) framework that learns from multimodal, heterogeneous datasets to improve clinical outcome prediction. PARADIGM generates embeddings from multi-resolution data using foundation models, aggregates them into patient-level representations, fuses them into a unified graph, and enhances performance for tasks like survival analysis. We train GNNs on pan-Squamous Cell Carcinomas and validate our approach on Moffitt Cancer Center lung SCC data. Multimodal GNN outperforms other models in patient survival prediction. Converging individual data modalities across varying scales provides a more insightful disease view. Our solution aims to understand the patient's circumstances comprehensively, offering insights on heterogeneous data integration and the benefits of converging maximum data views.

q-bio.CB↗

arXiv2024

Self-Normalizing Foundation Model for Enhanced Multi-Omics Data Analysis in Oncology

Multi-omics research has enhanced our understanding of cancer heterogeneity and progression. Investigating molecular data through multi-omics approaches is crucial for unraveling the complex biological mechanisms underlying cancer, thereby enabling more effective diagnosis, treatment, and prevention strategies. However, predicting patient outcomes through the integration of all available multi-omics data is still an under-study research direction. Here, we present SeNMo, a foundation model that has been trained on multi-omics data across 33 cancer types. SeNMo is particularly efficient in handling multi-omics data characterized by high-width and low-length attributes. We trained SeNMo for the task of overall survival of patients using pan-cancer multi-omics data involving 33 cancer sites from the GDC. The training multi-omics data includes gene expression, DNA methylation, miRNA expression, DNA mutations, protein expression modalities, and clinical data. SeNMo was validated on two independent cohorts: Moffitt Cancer Center and CPTAC lung squamous cell carcinoma. We evaluated the model's performance in predicting patient's overall survival using the C-Index. SeNMo performed consistently well in the training regime, reflected by the validation C-Index of 0.76 on GDC's public data. In the testing regime, SeNMo performed with a C-Index of 0.758 on a held-out test set. The model showed an average accuracy of 99.8% on the task of classifying the primary cancer type on the pan-cancer test cohort. SeNMo demonstrated robust performance on the classification task of predicting the primary cancer type of patients. SeNMo further demonstrated significant performance in predicting tertiary lymph structures from multi-omics data, showing generalizability across cancer types, molecular data types, and clinical endpoints.

cs.LG↗

arXiv2023

Multimodal Data Integration for Oncology in the Era of Deep Neural Networks: A Review

Cancer has relational information residing at varying scales, modalities, and resolutions of the acquired data, such as radiology, pathology, genomics, proteomics, and clinical records. Integrating diverse data types can improve the accuracy and reliability of cancer diagnosis and treatment. There can be disease-related information that is too subtle for humans or existing technological tools to discern visually. Traditional methods typically focus on partial or unimodal information about biological systems at individual scales and fail to encapsulate the complete spectrum of the heterogeneous nature of data. Deep neural networks have facilitated the development of sophisticated multimodal data fusion approaches that can extract and integrate relevant information from multiple sources. Recent deep learning frameworks such as Graph Neural Networks (GNNs) and Transformers have shown remarkable success in multimodal learning. This review article provides an in-depth analysis of the state-of-the-art in GNNs and Transformers for multimodal data fusion in oncology settings, highlighting notable research studies and their findings. We also discuss the foundations of multimodal learning, inherent challenges, and opportunities for integrative learning in oncology. By examining the current state and potential future developments of multimodal data integration in oncology, we aim to demonstrate the promising role that multimodal neural networks can play in cancer prevention, early detection, and treatment through informed oncology practices in personalized settings.

cs.LG↗

arXiv2015

The VAMPIRES instrument: Imaging the innermost regions of protoplanetary disks with polarimetric interferometry

Direct imaging of protoplanetary disks promises to provide key insight into the complex sequence of processes by which planets are formed. However imaging the innermost region of such disks (a zone critical to planet formation) is challenging for traditional observational techniques (such as near-IR imaging and coronagraphy) due to the relatively long wavelengths involved and the area occulted by the coronagraphic mask. Here we introduce a new instrument -- VAMPIRES -- which combines non-redundant aperture-masking interferometry with differential polarimetry to directly image this previously inaccessible innermost region. By using the polarisation of light scattered by dust in the disk to provide precise differential calibration of interferometric visibilities and closure phases, VAMPIRES allows direct imaging at and beyond the telescope diffraction limit. Integrated into the SCExAO system at the Subaru telescope, VAMPIRES operates at visible wavelengths (where polarisation is high) while allowing simultaneous infrared observations conducted by HICIAO. Here we describe the instrumental design and unique observing technique and present the results of the first on-sky commissioning observations, validating the excellent visibility and closure phase precision which are then used to project expected science performance metrics.

astro-ph.IM↗

arXiv2013

SCExAO as a precursor to an ELT exoplanet direct imaging instrument

The Subaru Coronagraphic Extreme AO (SCExAO) instrument consists of a high performance Phase Induced Amplitude Apodisation (PIAA) coronagraph combined with an extreme Adaptive Optics (AO) system operating in the near-infrared (H band). The extreme AO system driven by the 2000 element deformable mirror will allow for Strehl ratios >90% to be achieved in the H-band when it goes closed loop. This makes the SCExAO instrument a powerful platform for high contrast imaging down to angular separations of the order of 1lambda/D and an ideal testbed for exploring coronagraphic techniques for ELTs. In this paper we report on the recent progress in regards to the development of the instrument, which includes the addition of a visible bench that makes use of the light at shorter wavelengths not currently utilized by SCExAO and closing the loop on the tip/tilt wavefront sensor. We will also discuss several exciting guest instruments which will expand the capabilities of SCExAO over the next few years; namely CHARIS which is a integral field spectrograph as well as VAMPIRES, a visible aperture masking experiment based on polarimetric analysis of circumstellar disks. In addition we will elucidate the unique role extreme AO systems will play in enabling high precision radial velocity spectroscopy for the detection of small companions.

astro-ph.IM↗

arXiv2012

Design of Optically Path Length Matched, Three-Dimensional Photonic Circuits Comprising Uniquely Routed Waveguides

A method for designing physically path length matched, three-dimensional photonic circuits is described. We focus specifically on the case where all the waveguides are uniquely routed from the input to output; a problem which has not been addressed to date and allows for the waveguides to be used in interferometric measurements. Circuit elements were fabricated via the femtosecond laser direct-write technique. We demonstrate via interferometric methods that the fabricated circuits were indeed optically path length matched to within 45 um which is within the coherence length required for many applications.

astro-ph.IM↗

arXiv2012

A close halo of large transparent grains around extreme red giant stars

Intermediate-mass stars end their lives by ejecting the bulk of their envelope via a slow dense wind back into the interstellar medium, to form the next generation of stars and planets. Stellar pulsations are thought to elevate gas to an altitude cool enough for the condensation of dust, which is then accelerated by radiation pressure from starlight, entraining the gas and driving the wind. However accounting for the mass loss has been a problem due to the difficulty in observing tenuous gas and dust tens of milliarcseconds from the star, and there is accordingly no consensus on the way sufficient momentum is transferred from the starlight to the outflow. Here, we present spatially-resolved, multi-wavelength observations of circumstellar dust shells of three stars on the asymptotic giant branch of the HR diagram. When imaged in scattered light, dust shells were found at remarkably small radii (<~ 2 stellar radii) and with unexpectedly large grains (~300 nm radius). This proximity to the photosphere argues for dust species that are transparent to starlight and therefore resistant to sublimation by the intense radiation field. While transparency usually implies insufficient radiative pressure to drive a wind, the radiation field can accelerate these large grains via photon scattering rather than absorption - a plausible mass-loss mechanism for lower-amplitude pulsating stars.

astro-ph.SR↗

arXiv2011

Mapping the Shores of the Brown Dwarf Desert III: Young Moving Groups

We present the results of an aperture masking interferometry survey for substellar companions around 67 members of the young (~8-200Myr) nearby (~5-86pc) AB Doradus, Beta Pictoris, Hercules-Lyra, TW Hya, and Tucana-Horologium stellar associations. Observations were made at near infrared wavelengths between 1.2-3.8 microns using the adaptive optics facilities of the Keck II, VLT UT4, and Palomar Hale Telescopes. Typical contrast ratios of ~100-200 were achieved at angular separations between ~40-320mas, with our survey being 100% complete for companions with masses below 0.25\msolar across this range. We report the discovery of a $0.52 \pm 0.09$\msolar companion to HIP14807, as well as the detections and orbits of previously known stellar companions to HD16760, HD113449, and HD160934. We show that the companion to HD16760 is in a face-on orbit, resulting in an upward revision of its mass from $M_2 \sin i \sim 14$\mjupiter to $M_2 = 0.28 \pm 0.04$\msolar. No substellar companions were detected around any of our sample members, despite our ability to detect companions with masses below 80\mjupiter for 50 of our targets: of these, our sensitivity extended down to 40\mjupiter around 30 targets, with a subset of 22 subject to the still more stringent limit of 20\mjupiter. A statistical analysis of our non-detection of substellar companions allows us to place constraints on their frequency around ~0.2-1.5\msolar stars. In particular, considering companion mass distributions that have been proposed in the literature, we obtain an upper limit estimate of ~9-11% for the frequency of 20-80\mjupiter companions between 3-30AU at 95% confidence, assuming that their semimajor axes are distributed according to $d\mathcal{N}/da \propto a^{-1}$ in this range.

astro-ph.SR↗

arXiv2010

Photonic Technologies for a Pupil Remapping Interferometer

Interest in pupil-remapping interferometry, in which a single telescope pupil is fragmented and recombined using fiber optic technologies, has been growing among a number of groups. As a logical extrapolation from several highly successful aperture masking programs underway worldwide, pupil remapping offers the advantage of spatial filtering (with single-mode fibers) and in principle can avoid the penalty of low throughput inherent to an aperture mask. However in practice, pupil remapping presents a number of difficult technological challenges including injection into the fibers, pathlength matching of the device, and stability and reproducibility of the results. Here we present new approaches based on recently-available photonic technologies in which coherent three-dimensional waveguide structures can be sculpted into bulk substrate. These advances allow us to miniaturize the photonic processing into a single, robust, thermally stable element; ideal for demanding observatory or spacecraft environments. Ultimately, a wide range of optical functionality could be routinely fabricated into such structures, including beam combiners and dispersive or wavelength selective elements, bringing us closer to the vision of an interferometer on a chip.

astro-ph.IM↗

arXiv2010

Sparse Aperture Masking (SAM) at NAOS/CONICA on the VLT

The new operational mode of aperture masking interferometry has been added to the CONICA camera which lies downstream of the Adaptive Optics (AO) corrected focus provided by NAOS on the VLT-UT4 telescope. Masking has been shown to deliver superior PSF calibration, rejection of atmospheric noise and robust recovery of phase information through the use of closure phases. Over the resolution range from about half to several resolution elements, masking interferometry is presently unsurpassed in delivering high fidelity imaging and direct detection of faint companions. Here we present results from commissioning data using this powerful new operational mode, and discuss the utility for masking in a variety of scientific contexts. Of particular interest is the combination of the CONICA polarimetry capabilities together with SAM mode operation, which has revealed structures never seen before in the immediate circumstellar environments of dusty evolved stars.

astro-ph.IM↗