Biology subjects

Ganjgahi, H.

Publications and source records attributed to Ganjgahi, H..

8 recordsLinked to original sources

bioRxiv2026

ComCat: Combating Covariate Effects in Brain Analysis

As neuroimaging analysis shifts toward large-scale, multi-site studies, managing the unwanted variability introduced by combining heterogeneous datasets has become a critical challenge. Although tools such as ComBat and its neuroimaging extensions are widely used to address this variability, they only permit the modeling of categorical site effects and cannot account for continuous sources of confounding, such as image quality, head motion, and acquisition parameters. We introduce ComCat, an extension of the ComBat framework that preserves biologically relevant covariates while removing the effects of categorical site indicators and continuous nuisance variables. The latter are modeled as smooth nonlinear functions via B-spline basis expansion. ComCat is applicable to a broad range of brain analysis tasks, including voxel- and surface-based morphometry, normative modeling, and machine learning-based prediction. To demonstrate its capabilities, we evaluated ComCat on brain age prediction across five datasets covering complementary multi-site harmonization scenarios: ON-Harmony (10 subjects x 6 scanners; n = 80); the Buchert traveling-phantom dataset (1 subject x 116 scanners; n = 531); the Tohoku single-scanner, varying-acquisition dataset (n = 121); MR-ART (148 subjects with varying motion levels); and an ABIDE subset comprising 229 control subjects and 208 individuals with autism spectrum disorder across 14 scanners. Using image quality measures derived from CAT12 as continuous nuisance variables, ComCat reduced the mean absolute error (MAE) in brain age prediction relative to ComBat-GAM in all five datasets, including the two scenarios where site information was unavailable or uninformative. In the ABIDE dataset, ComCat improved harmonization while preserving the difference between the control and ASD groups, demonstrating that scanner-related variance can be removed without affecting biologically meaningful signals. ComCat can operate with or without site labels and is agnostic to the source of image quality metrics.

neuroscience↗

bioRxiv2026

MIDFA: Scalable Bayesian Factor Analysis for Mixed and Incomplete Data

Probabilistic latent variable models are a powerful tool for uncovering structure in high-dimensional datasets, particularly in biomedical applications. The increasing availability of large-scale epidemiological studies, such as the UK Biobank, poses important modelling challenges, including mixed data types, high dimensionality, and structured missingness. Existing approaches address some of these issues, but few provide a unified and scalable framework for handling them simultaneously. Here, we propose a scalable Bayesian factor analysis framework designed to address these challenges. Our method combines a semi-parametric Gaussian copula model with a continuous spike-and-slab prior to induce sparse and interpretable factor loadings. The number of latent dimensions is learned nonparametrically from the data using an Indian buffet process prior. For model fitting, we develop an expectation-maximisation algorithm that naturally accommodates missing data. We validate the proposed method through comprehensive simulation studies. In addition, we showcase the proposed model using the Novartis-Oxford Multiple Sclerosis dataset in two ways. First, we identify latent dimensions shared across MS clinical and neuroimaging variables, characterising disease structure while demonstrating the models ability to handle multiple data types and structured missingness. Second, we use the model for dimensionality reduction of structural MRI data, extracting features for downstream analysis that go beyond traditional whole-brain summary statistics. In those applications, our method identifies sparse latent structures and provides insights beyond those obtained from traditional approaches.

neuroscience↗

bioRxiv2025

BARTharm: MRI Harmonization Using Image Quality Metrics and Bayesian Non-parametric

Image derived phenotypes (IDPs) harmonization from Magnetic Resonance Imaging (MRI) data is essential for reducing scanner-induced, non-biological variability and enabling accurate multi-site analysis. Existing methods like ComBat, while widely used, rely on linear assumptions and explicit scanner IDs - limitations that reduce their effectiveness in real-world scenarios involving complex scanner effects, non-linear biological variation, or anonymized data. We introduce BARTharm, a novel harmonization framework that uses Image Quality Metrics (IQMs) instead of Scanner IDs and models scanner and biological effects separately using Bayesian Additive Regression Trees (BART), allowing for flexible, data-driven adjustment of IDPs. Through extensive simulation studies, we demonstrate that IQMs provide a more informative and flexible representation of scanner-related variation than categorical Scanner IDs, enabling more accurate removal of non-biological effects. Leveraging this and its ability to model complex relationships, BARTharm, consistently outperforms ComBat across a range of challenging scenarios, including model misspecification and confounded scanner-biological relationships. Applied to real-world datasets, BARTharm successfully removes scanner-induced bias while preserving meaningful biological signals, resulting in stronger, more reliable associations with clinical outcomes. Overall, we find that BARTharm is a robust, data-driven improvement over traditional harmonization approaches, particularly suited for modern, large-scale neuroimaging studies.

neuroscience↗

bioRxiv2022

A scalable approach for continuous time Markov models with covariates

Existing methods for fitting continuous time Markov models (CTMM) in the presence of covariates suffer from scalability issues due to high computational cost of matrix exponentials calculated for each observation. In this paper we propose an optimization technique for CTMM which uses a stochastic gradient descent algorithm combined with differentiation of the matrix exponential using a Pade approximation which we show then makes fitting large scale data feasible. We compare two methods for computing standard errors, one using Pade expansion, and the other using power series expansion of the matrix exponential. Through simulations we find improved performance relative to existing CTMM methods, and we demonstrate the method on the large-scale multiple sclerosis NO.MS dataset.

bioinformatics↗

bioRxiv2021

A convolutional neural network for the quality control of MRI defacing

Large-scale neuroimaging datasets present unique challenges for automated processing pipelines. Mo-tivated by a large clinical trials dataset of Multiple Sclerosis with over 235,000 MRI scans, we consider the challenge of defacing - anonymisation to remove identifying facial features. The defacing process must undergo quality control (QC) checks to ensure that the facial features have been adequately anonymised and that the brain tissue is left completely intact. Visual QC checks - particularly on a project of this scale - are time-consuming and can cause delays in preparing data. We have developed a convolutional neural network (CNN) that can assist with the QC of the application of MRI defacing; our CNN is able to distinguish between scans that are correctly defaced, and can also classify defacing failures into three sub-types to facilitate parameter tuning during remedial re-defacing. Since integrating the CNN into our anonymisation pipeline, over 75,000 scans have been processed. Strict thresholds have been applied so that ambiguous classifications are referred for visual QC checks, however all scans still undergo a time-efficient verification check before being marked as passed. After applying the thresholds, our network is 92% accurate and is able to classify nearly half of the scans without the need for protracted manual checks. Even with the introduction of the verification checks, incorporation of the CNN has reduced the time spent undertaking QC checks by 42% during initial defacing, and by 35% overall. With the help of the CNN, we have been able to successfully deface 96% of the scans in the project whilst maintaining high QC standards. In a similarly sized new project of this scale, we would expect the model to reduce the time spent on manual QC checks by 125 hours. Our approach is applicable to other projects with the potential to greatly improve the efficiency of imaging anonymisation pipelines.

neuroscience↗

bioRxiv2017

Fast and Powerful Genome Wide Association Analysis of Dense Genetic Data with High Dimensional Imaging Phenotypes

Genome wide association (GWA) analysis of brain imaging phenotypes can advance our understanding of the genetic basis of normal and disorder-related variation in the brain. GWA approaches typically use linear mixed effect models to account for non-independence amongst subjects due to factors such as family relatedness and population structure. The use of these models with high-dimensional imaging phenotypes presents enormous challenges in terms of computational intensity and the need to account multiple testing in both the imaging and genetic domain. Here we present method that makes mixed models practical with high-dimensional traits by a combination of a transformation applied to the data and model, and the use of a non-iterative variance component estimator. With such speed enhancements permutation tests are feasible, which allows inference on powerful spatial tests like the cluster size statistic.

genetics↗

bioRxiv2017

Do Candidate Genes Affect the Brain’s White Matter Microstructure? Large-Scale Evaluation of 6,165 Diffusion MRI Scans

AbstractSusceptibility genes for psychiatric and neurological disorders - including APOE, BDNF, CLU,CNTNAP2, COMT, DISC1, DTNBP1, ErbB4, HFE, NRG1, NTKR3, and ZNF804A - have been reported to affect white matter (WM) microstructure in the healthy human brain, as assessed through diffusion tensor imaging (DTI). However, effects of single nucleotide polymorphisms (SNPs) in these genes explain only a small fraction of the overall variance and are challenging to detect reliably in single cohort studies. To date, few studies have evaluated the reproducibility of these results. As part of the ENIGMA-DTI consortium, we pooled regional fractional anisotropy (FA) measures for 6,165 subjects (CEU ancestry N=4,458) from 11 cohorts worldwide to evaluate effects of 15 candidate SNPs by examining their associations with WM microstructure. Additive association tests were conducted for each SNP. We used several meta-analytic and mega-analytic designs, and we evaluated regions of interest at multiple granularity levels. The ENIGMA-DTI protocol was able to detect single-cohort findings as originally reported. Even so, in this very large sample, no significant associations remained after multiple-testing correction for the 15 SNPs investigated. Suggestive associations (1.3x10-4 < p < 0.05, uncorrected) were found for BDNF, COMT, and ZNF804A in specific tracts. Meta-and mega-analyses revealed similar findings. Regardless of the approach, the previously reported candidate SNPs did not show significant associations with WM microstructure in this largest genetic study of DTI to date; the negative findings are likely not due to insufficient power. Genome-wide studies, involving large-scale meta-analyses, may help to discover SNPs robustly influencing WM microstructure.

neuroscience↗

bioRxiv2016

fMRI Correlates of Reaction Time Prolongation during intentional False Responding; an inter-individual difference study

Reaction time (RT) is chiefly longer when people lie. However, the baseline speed in answering questions and the amount of RT prolongation during lying show considerable amount of inter-individual variability. In the current study, we exploited this fact to glean insights on the contribution of each lie-related brain region to hampering of response speeds when people try to be deceitful. In an event-related fMRI session, participants were interrogated by yes-no autobiographical questions and were instructed to intentionally provide false responses to a pre-selected subset of questions. Data from twenty healthy volunteers were analyzed. Baseline speed [RTtruth] and relative appended lie RT [(RTlie - RTtruth) / RTtruth] measures were calculated for each participant and were included in the group level analysis of [lie > truth] BOLD contrasts. Lying RTs were significantly longer than truth telling RTs. Lie-related increase in activity of right ventrolateral prefrontal cortex (VLPFC) and bilateral paracingulate cortex correlated with the baseline speed of participants, while the increase in activity of Left VLPFC, left lateral occipital cortex and bilateral anterior cingulate areas directly correlated with the amount of lying reaction time cost. Activity within bilateral posterior cingulate cortex and right insular cortex inversely correlated with lying RT-cost. Bilateral supplementary motor areas, internal capsule white matter and left angular gyrus showed lie-related increase in activity but did not correlate with either of behavioral measures. Provisional implications regarding the contribution of these regions to RT prolongation and their cognitive role in deceitful behavior are discussed.

neuroscience↗