Biology subjects

Perrone, A. J.

Publications and source records attributed to Perrone, A. J..

6 recordsLinked to original sources

bioRxiv2024

Baby Open Brains: An Open-Source Repository of Infant Brain Segmentations

Reproducibility of neuroimaging research on infant brain development remains limited due to highly variable protocols and processing approaches. Progress towards reproducible pipelines is limited by a lack of benchmarks such as gold standard brain segmentations. Addressing this core limitation, we constructed the Baby Open Brains (BOBs) Repository, an open source resource comprising manually curated and expert-reviewed infant brain segmentations. Markers and expert reviewers manually segmented anatomical MRI data from 71 infant imaging visits across 51 participants, using both T1w and T2w images per visit. Anatomical images showed dramatic differences in myelination and intensities across the 1 to 9 month age range, emphasizing the need for densely sampled gold standard manual segmentations in these ages. The BOBs repository is publicly available through the Masonic Institute for the Developing Brain (MIDB) Open Data Initiative, which links S3 storage, Datalad for version control, and BrainBox for visualization. This repository represents an open-source paradigm, where new additions and changes can be added, enabling a community-driven resource that will improve over time and extend into new ages and protocols. These manual segmentations and the ongoing repository provide a benchmark for evaluating and improving pipelines dependent upon segmentations in the youngest populations. As such, this repository provides a vitally needed foundation for early-life large-scale studies such as HBCD.

neuroscience↗

bioRxiv2024

Heritability estimation of subcortical volumes in a multi-ethnic multi-site cohort study

Heritability of regional subcortical brain volumes (rSBVs) describes the role of genetics in middle and inner brain development. rSBVs are highly heritable in adults but are not characterized well in adolescents. The Adolescent Brain Cognitive Development study (ABCD), taken over 22 US sites, provides data to characterize the heritability of subcortical structures in adolescence. In ABCD, site-specific effects co-occur with genetic effects which can bias heritability estimates. Existing methods adjusting for site effects require additional steps to adjust for site effects and can lead to inconsistent estimation. We propose a random-effect model-based method of moments approach that is a single step estimator and is a theoretically consistent estimator even when sites are imbalanced and performs well under simulations. We compare methods on rSBVs from ABCD. The proposed approach yielded heritability estimates similar to previous results derived from single-site studies. The cerebellum cortex and hippocampus were the most heritable regions (>50%).

genomics↗

bioRxiv2023

BIBSNet: A Deep Learning Baby Image Brain Segmentation Network for MRI Scans

ObjectivesBrain segmentation of infant magnetic resonance (MR) images is vitally important for studying typical and atypical brain development. The infant brain undergoes many changes throughout the first years of postnatal life, making tissue segmentation difficult for most existing algorithms. Here we introduce a deep neural network BIBSNet (Baby and Infant Brain Segmentation Neural Network), an open-source, community-driven model for robust and generalizable brain segmentation leveraging data augmentation and a large sample size of manually annotated images. Experimental DesignIncluded in model training and testing were MR brain images from 90 participants with an age range of 0-8 months (median age 4.6 months). Using the BOBs repository of manually annotated real images along with synthetic segmentation images produced using SynthSeg, the model was trained using a 10-fold procedure. Model performance of segmentations was assessed by comparing BIBSNet, joint label fusion (JLF) inferred segmentation to ground truth segmentations using Dice Similarity Coefficient (DSC). Additionally, MR data along with the FreeSurfer compatible segmentations were processed with the DCAN labs infant-ABCD-BIDS processing pipeline from ground truth, JLF, and BIBSNet to further assess model performance on derivative data, including cortical thickness, resting state connectivity and brain region volumes. Principal ObservationsBIBSNet segmentations outperforms JLF across all regions based on DSC comparisons. Additionally, with processed derived metrics, BIBSNet segmentations outperforms JLF segmentations across nearly all metrics. ConclusionsBIBSNet segmentation shows marked improvement over JLF across all age groups analyzed. The BIBSNet model is 600x faster compared to JLF, produces FreeSurfer-compatible segmentation labels, and can be easily included in other processing pipelines. BIBSNet provides a viable alternative for segmenting the brain in the earliest stages of development.

neuroscience↗

bioRxiv2022

Motion Impact Score for Detecting Spurious Brain-Behavior Associations

Between-participant differences in head motion introduce systematic bias to resting state fMRI brain-wide association studies (BWAS) that is not completely removed by denoising algorithms. Researchers who study traits, or phenotypes associated with in-scanner head motion (e.g. psychiatric disorders) need to know if trait-functional connectivity (FC) effects are biased by residual motion artifact in order to avoid reporting false positive results. We devised an adaptable method, Split Half Analysis of Motion Associated Networks (SHAMAN), to assign a motion impact score to specific trait-FC effects. The SHAMAN approach distinguishes between motion artifact causing overestimation or underestimation of trait-FC effects. SHAMAN was > 95% specific at sample sizes of n = 100 and above. SHAMAN was powered to detect motion overestimation scores 80% of the time at sample sizes of n = 5,000 but could detect motion underestimation scores only 50% of the time at n = 5000, making it most useful for researchers seeking to avoid overestimating trait-FC effects in large BWAS. We computed motion impact scores for trait-FC effect with 45 demographic, biophysical, cognitive, and personality traits from n = 7,270 participants in the Adolescent Brain Cognitive Development (ABCD) Study. After standard denoising with ABCD-BIDS and without motion censoring, 42% (19/45) of traits had significant (p < 0.05) motion overestimation scores and 38% (17/45) of traits had significant motion underestimation scores. Censoring at framewise displacement (FD) < 0.2 mm reduced the proportion of traits with significant motion overestimation scores from 42% to 2% (1/45) but did not decrease the number of traits with significant motion underestimation scores.

neuroscience↗

bioRxiv2022

Curation of BIDS (CuBIDS): a workflow and software package for streamlining reproducible curation of large BIDS datasets

The Brain Imaging Data Structure (BIDS) is a specification accompanied by a software ecosystem that was designed to create reproducible and automated workflows for processing neuroimaging data. BIDS Apps flexibly build workflows based on the metadata detected in a dataset. However, even BIDS valid metadata can include incorrect values or omissions that result in inconsistent processing across sessions. Additionally, in large-scale, heterogeneous neuroimaging datasets, hidden variability in metadata is difficult to detect and classify. To address these challenges, we created a Python-based software package titled "Curation of BIDS" (CuBIDS), which provides an intuitive workflow that helps users validate and manage the curation of their neuroimaging datasets. CuBIDS includes a robust implementation of BIDS validation that scales to large samples and incorporates DataLad--a version control software package for data--to ensure reproducibility and provenance tracking throughout the entire curation process. CuBIDS provides tools to help users perform quality control on their images metadata and identify unique combinations of imaging parameters. Users can then execute BIDS Apps on a subset of participants that represent the full range of acquisition parameters that are present, accelerating pipeline testing on large datasets. HIGHLIGHTSO_LICuBIDS is a workflow and software package for curating BIDS data. C_LIO_LICuBIDS summarizes the heterogeneity in a BIDS dataset. C_LIO_LICuBIDS prepares BIDS data for successful preprocessing pipeline runs. C_LIO_LICuBIDS helps users perform metadata-based quality control. C_LI

neuroscience↗

bioRxiv2021

Using synthetic MR images for field map-less distortion correction

Functional MRI (fMRI) data acquired using echo-planar imaging (EPI) are highly distorted by magnetic field inhomogeneities. Distortion combined with underlying differences in image contrast between EPI and T1-weighted and T2-weighted (T1w/T2w) structural images makes the alignment of functional and anatomical images a challenge. Typically, separately acquired field map data are used to correct fMRI distortions and a flexible cost function insensitive to cross-modal differences in image contrast and intensity is used for aligning fMRI and anatomical images. The quality of alignment achieved with this approach can vary greatly and depends on the quality of field map data. In addition, many publicly available datasets lack field map data entirely. To address this issue, we developed Synth, a software package for distortion correction and cross-modal image registration that does not require separately acquired field map data. Synth combines information from T1w and T2w anatomical images to construct an idealized undistorted synthetic image that has similar contrast properties to fMRI data. The undistorted synthetic image then serves as an effective reference for individual-specific nonlinear unwarping to correct fMRI distortions. We demonstrate, in both pediatric (ABCD: Adolescent Brain Cognitive Development) and adult (MSC: Midnight Scan Club) data that Synth performs comparably well to other leading distortion correction approaches that utilize field map data, and often outperforms them. Field map-less distortion correction with Synth allows accurate and precise registration of fMRI data with missing or corrupted field map information.

neuroscience↗