Biology subjects

Oberai, A.

Publications and source records attributed to Oberai, A..

5 recordsLinked to original sources

bioRxiv2025

Probabilistic Multi-site MR Image Harmonization via Feature Preserving Conditional Generative Adversarial Networks

Brain magnetic resonance imaging (MRI) is pivotal in diagnosing and monitoring neurological disorders. However, despite their extensive applications, MR images have certain shortcomings. In particular, factors other than the anatomy of brain tissues influence the intensity distribution of voxels in MR images. These factors include hardware, software, magnetic field strength, and acquisition protocol. This inconsistency poses challenges in multi-site neuroimaging studies, where images are obtained from various devices with minimal control over acquisition parameters. Image harmonization algorithms aim to eliminate non-biological characteristics in MR images through various approaches, including converting images from multiple sites into a format resembling that of a designated target site. Among image harmonization methods, those relying on deep learning algorithms have gained significant attention recently. Nevertheless, certain aspects of deep learning-based image harmonization remain unexplored, notably the integration of probabilistic deep generative models to transform the distribution of MR images to a desired distribution. Inspired by this, we introduced a feature preserving conditional generative adversarial network (FP-cGAN) that converts images from multiple origins into the format of a target site while preserving anatomical features by imposing a novel regularizing constraint. We conduct our experiments on MR images from the SRPBS dataset, which comprises unpaired images in addition to paired (traveling subjects) images from multiple sites. We utilize the unpaired data for training our models and the paired data for evaluation. Furthermore, we compare our results with cycleGAN and histogram matching, two widely used image harmonization methods. Our experiments reveal that our approach surpasses the other techniques. Article HighlightsO_LIWe introduce an FP-cGAN model to harmonize multi-site MRI scans into a target site. C_LIO_LIA novel anatomical feature invariant constraint maintains tissue integrity. C_LIO_LIThe models are trained in an unpaired setting without requiring traveling subjects. C_LIO_LIWe evaluate harmonization performance using paired traveling subjects. C_LI

bioengineering↗

bioRxiv2025

Contrastive Representation Learning for Single Cell Phenotyping in Whole Slide Imaging of Enrichment-free Liquid Biopsy

Tumor-associated cells derived from a liquid biopsy are promising biomarkers for cancer detection, diagnosis, prognosis, and monitoring. However, their rarity, heterogeneity and plasticity make precise identification and biological characterization challenging for clinical utility. Enrichment-free approaches using whole slide imaging of all circulating cells offer a comprehensive and unbiased strategy for capturing the full spectrum of tumor-associated cell phenotypes. However, current analysis methods often depend on engineered features and manual expert review, making them sensitive to technical variations and subjective biases. These limitations highlight the need for a better feature representation to improve performance and reproducibility of applications in large-scale patient cohort analyses. In this study, we present a deep contrastive learning framework for learning features of all circulating cells, enabling robust identification and stratification of single cells in whole slide immunofluorescence microscopy images. We demonstrate performance of learned features in classification of diverse cell phenotypes in the liquid biopsy, achieving an accuracy of 92.64%. We further demonstrate that learned features improve performance in downstream applications such as outlier detection and clustering. Lastly, our feature representation enables automated identification and enumeration of distinct rare cell phenotypes, achieving average F1-score of 0.93 across cell lines mimicking circulating tumor cells and endothelial cells in contrived samples and average F1-score of 0.858 across CTC phenotypes in clinical samples. This workflow has significant implications for scalable analysis of tumor-associated cellular biomarkers in clinical prognosis and personalized treatment strategies.

cancer biology↗

bioRxiv2024

Probabilistic Brain MR Image Transformation Using Generative Models

Brain MR image transformation, which is the process of transforming one type of MR image into another, is a critical neuroimaging task that is needed when the target image type is missing or corrupted. Accordingly, several methods have been developed to tackle this problem, with a recent focus on deep learning-based models. In this paper, we investigate the performance of the conditional version of three such probabilistic generative models, including conditional Generative Adversarial Networks (cGAN), Noise Conditioned Score Networks (NCSN), and De-noising Diffusion Probabilistic Models (DDPM). We also compare their performance against a more traditional deterministic U-Net based model. We train and test these models using MR images from publicly available datasets IXI and OASIS. For images from the IXI dataset, we conduct experiments on combinations of transformations between T1-weighted (T1), T2-weighted (T2), and proton density (PD) images, whereas for the OASIS dataset, we consider combinations of T1, T2, and Fluid Attenuated Inversion Recovery (FLAIR) images. In evaluating these models, we measure the similarity between the transformed image and the target image using metrics like PSNR and SSIM. In addition, for the three probabilistic generative models, we evaluate the utility of generating an ensemble of predictions by computing a metric that measures the variance in their predictions and demonstrate that it can be used to identify out-of-distribution (OOD) input images. We conclude that the NCSN model yields the most accurate transformations, while the DDPM model yields variance results that most clearly detect OOD inputs. We also note that while the results for the two diffusion models (NCSN and DDPM) are more accurate than those for the cGAN, the latter was significantly more efficient in generating multiple samples. Overall, our work demonstrates the utility of probabilistic conditional generative models for MR image transformation and highlights the role of generating an ensemble of outputs in identifying OOD input images.

bioengineering↗

bioRxiv2024

A Novel Pan-Proteome Array for High-Throughput Profiling of the Humoral Response to Treponema pallidum subsp. pallidum: a Pre-Clinical Study

BackgroundGiven the resurgence of syphilis, research endeavors to improve current assays for serological diagnosis and management of this disease are a priority. A proteome-scale platform for high-throughput profiling of the humoral response to Treponema pallidum (T. pallidum) proteins during infection could identify antigens suitable to ameliorate the performance and capabilities of treponemal tests (TTs), which may require weeks to become positive following infection, cannot distinguish between active and previously treated infections, or assess treatment response. Additionally, because infection-induced immunity is partially protective, profiling the response to T. pallidum outer membrane proteins (OMPs) could help select vaccine candidates. MethodsWe developed a pan-proteome array (PPA) based on the Nichols and SS14 strain complete proteomes and used it to define the IgM and IgG humoral response to 1,009 T. pallidum proteins in sera collected longitudinally from long-term infected rabbits, and from rabbits that were infected, treated, and re-infected. FindingsApproximately a third of the pathogens proteome was recognized in infected animals, with a marked IgG response detectable between day-10 and day-20 post-infection. We found early, gradual, and late IgG kinetic profiles, strain-dependent differences in humoral reactivity, and post-treatment fluctuation in reactivity for several antigens. Very few antigens elicited an IgM response. Several OMPs were significantly and differentially recognized, but few elicited a robust response. InterpretationThe PPA allowed the identification of antigens that could facilitate early diagnosis and of a core set of OMP that could explain protection upon re-infection. No antigen appeared suitable to monitor treatment response. FundingNIH SBIR-R43AI149804 RESEARCH IN CONTEXTO_ST_ABSEvidence before this studyC_ST_ABSIn April 2024, we searched the PubMed database for articles on preclinical studies using high throughput proteome arrays containing at least 10% of the predicted T. pallidum proteome that aimed at identifying antibody reactivity to T. pallidum antigens during experimental syphilis infection. We could retrieve only one manuscript. In this work, an array containing the T. pallidum partial proteome as annotated in the first sequenced Nichols strain genome (GCA_000008605.1) in 1998 was assembled using recombinant antigens expressed in Escherichia coli (E. coli). The resulting array was probed using pooled sera from three rabbits infected with the Nichols stain of T. pallidum, attained from infected animals at five time points following intratesticular infection. The small number of reactive antigens (n = 106) identified in this early study was likely to be an incomplete set of all antigens recognized during infection because not all the predicted targets in the T. pallidum proteome were successfully expressed and tested. In retrospect, additional limitations of the study included an initial suboptimal annotation of the Nichols genome used to define the pathogens proteome, which has now changed with the availability of a re-sequenced Nichols strain genome devoid of sequencing errors that affected the initial annotation process, and the refinement of bioinformatic pipelines for the identification of open reading frames (ORFs). Furthermore (as acknowledged by the authors), the possible presence of amplification errors in their expression clones might have affected the sequence of some protein targets and antibody binding to the targets. As a result, some of the T. pallidum antigens known to elicit a robust humoral response during experimental infection were not detected in this antigenic screen. Lastly, employing only the Nichols strain in this early study did not consider that a significant portion of the circulating syphilis strains belong to the SS14 clade of T. pallidum. Added value of this studyThis novel PPA, combined with a more robust experiential design than ever reported, allowed us to overcome most of the limitations associated with the study mentioned above, as we were able to a) use the most recent annotations for the selected T. pallidum strains based on accurate genome sequences, b) print the pathogens virtually complete proteome in the study array, c) analyze individual sera to account for rabbit-to-rabbit variability in the humoral response to infection rather than pooled sera, d) detect both IgM and IgG over 10 or 20 timepoints, depending on the experimental design, e) obtain information on how the humoral response evolved upon treatment and re-infection and, finally, f) evaluate all of the above in animals infected with two T. pallidum strains whose genetic background is representative of the two currently circulating clades of the syphilis agent. Implications of all the available evidenceOur study provides new and more comprehensive data on how humoral immunity for two classes of antibodies develops during infection and how it evolves in response to treatment and re-infection. The analysis of sera collected at tightly spaced time points post-inoculation and for an extensive period post-infection provides a wealth of information to improve the diagnostic performance of existing tests detecting treponemal antigens. The analysis of differential immunity specific to the pathogens putative OMPs provides a rationale for vaccine candidate selection.

microbiology↗

bioRxiv2022

A study of hyperelastic continuum models for isotropic athermal fibrous networks

Many biological materials contain fibrous protein networks as their main structural components. Understanding the mechanical properties of such networks is important for creating biomimicking materials for cell and tissue engineering, and for developing novel tools for detecting and diagnosing disease. In this work, we develop continuum models for isotropic, athermal fibrous networks by combining a single-fiber model that describes the axial response of individual fibers, with network models that assemble individual fiber properties into overall network behavior. In particular, we consider four different network models, including the affine, three-chain, eight-chain, and micro-sphere models, which employ different assumptions about network structure and kinematics. We systematically investigate the ability of these models to describe the mechanical response of athermal collagen and fibrin networks by comparing model predictions with experimental data. We test how each model captures network behavior under three different loading conditions: uniaxial tension, simple shear, and combined tension and shear. We find that the affine and three-chain models can accurately describe both the axial and shear behavior, whereas the eight-chain and micro-sphere models fail to capture the shear response, leading to an unphysical zero shear moduli at infinitesimal strains. Our study is the first to systematically investigate the applicability of popular network models for describing the macroscopic behavior of athermal fibrous networks, offering insights for selecting efficient models that can be used for large-scale, finite-element simulations of athermal networks.

biophysics↗