FastSurferVINN: Building resolution-independence into deep learning segmentation methods—A solution for HighRes brain MRI

NeuroImage, Год журнала: 2022, Номер 251, С. 118933 - 118933

Опубликована: Фев. 3, 2022

Leading neuroimaging studies have pushed 3T MRI acquisition resolutions below 1.0 mm for improved structure definition and morphometry. Yet, only few, time-intensive automated image analysis pipelines been validated high-resolution (HiRes) settings. Efficient deep learning approaches, on the other hand, rarely support more than one fixed resolution (usually mm). Furthermore, lack of a standard submillimeter as well limited availability diverse HiRes data with sufficient coverage scanner, age, diseases, or genetic variance poses additional, unsolved challenges training networks. Incorporating resolution-independence into learning-based segmentation, i.e., ability to segment images at their native across range different voxel sizes, promises overcome these challenges, yet no such approach currently exists. We now fill this gap by introducing Voxelsize Independent Neural Network (VINN) resolution-independent segmentation tasks present FastSurferVINN, which (i) establishes implements first method simultaneously supporting 0.7-1.0 whole brain (ii) significantly outperforms state-of-the-art methods resolutions, (iii) mitigates imbalance problem in datasets. Overall, internal mutually benefits both segmentation. With our rigorously FastSurferVINN we distribute rapid tool morphometric neuroimage analysis. The VINN architecture, furthermore, represents an efficient wider application

Язык: Английский

---

Survey on Open Science Practices in Functional Neuroimaging

NeuroImage, Год журнала: 2022, Номер 257, С. 119306 - 119306

Опубликована: Май 17, 2022

Replicability and reproducibility of scientific findings is paramount for sustainable progress in neuroscience. Preregistration the hypotheses methods an empirical study before analysis, sharing primary research data, compliance with data standards such as Brain Imaging Data Structure (BIDS), are considered effective practices to secure substantiate quality research. We investigated current level adoption open science neuroimaging difficulties that prevent researchers from using them. Email invitations participate survey were sent addresses received through a PubMed search human functional magnetic resonance imaging studies published between 2010 2020. 283 persons completed questionnaire. Although half participants experienced preregistration, willingness preregister future was modest. The majority had experience data. Most interested implementing standardized structure BIDS their labs. Based on demographic variables, we compared seven subscales, which been generated factor analysis. Exploratory analyses found at lower career higher fear being transparent residence EU need governance. Additionally, medical faculties other university reported more unsupportive supervisor regards results suggest growing but also highlight number important impediments.

Язык: Английский

---

Large, open datasets for human connectomics research: Considerations for reproducible and responsible data use

NeuroImage, Год журнала: 2021, Номер 244, С. 118579 - 118579

Опубликована: Сен. 15, 2021

Large, open datasets have emerged as important resources in the field of human connectomics. In this review, evolution data sharing involving magnetic resonance imaging is described. A summary challenges and progress conducting reproducible analyses provided, including description recent made development community guidelines recommendations, software management tools, initiatives to enhance training education. Finally, review concludes with a discussion ethical conduct relevant large, researcher's responsibility prevent further stigmatization historically marginalized racial ethnic groups. Moving forward, future work should include an enhanced emphasis on social determinants health, which may contextualize findings among diverse population-based samples. Leveraging date guided by interdisciplinary collaborations, connectomics promises be impressive era innovative research, yielding more inclusive understanding brain structure function.

Язык: Английский

---

Machine learning techniques for the Schizophrenia diagnosis: a comprehensive review and future research directions

Journal of Ambient Intelligence and Humanized Computing, Год журнала: 2023, Номер 14(5), С. 4795 - 4807

Опубликована: Март 9, 2023

Язык: Английский

---

Towards Automated Brain Aneurysm Detection in TOF-MRA: Open Data, Weak Labels, and Anatomical Knowledge

Neuroinformatics, Год журнала: 2022, Номер 21(1), С. 21 - 34

Опубликована: Авг. 18, 2022

Abstract Brain aneurysm detection in Time-Of-Flight Magnetic Resonance Angiography (TOF-MRA) has undergone drastic improvements with the advent of Deep Learning (DL). However, performances supervised DL models heavily rely on quantity labeled samples, which are extremely costly to obtain. Here, we present a model for that overcomes issue “weak” labels: oversized annotations considerably faster create. Our weak labels resulted be four times generate than their voxel-wise counterparts. In addition, our leverages prior anatomical knowledge by focusing only plausible locations occurrence. We first train and evaluate through cross-validation an in-house TOF-MRA dataset comprising 284 subjects (170 females / 127 healthy controls 157 patients 198 aneurysms). On this dataset, best achieved sensitivity 83%, False Positive (FP) rate 0.8 per patient. To assess generalizability, then participated challenge data (93 patients, 20 controls, 125 public challenge, was 68% (FP = 2.5), ranking 4th/18 open leaderboard. found no significant difference between risk-of-rupture groups ( p 0.75), 0.72), or sizes 0.15). Data, code weights released under permissive licenses. demonstrate can alleviate necessity prohibitively expensive annotations.

Язык: Английский

---

qMRI-BIDS: An extension to the brain imaging data structure for quantitative magnetic resonance imaging data

Scientific Data, Год журнала: 2022, Номер 9(1)

Опубликована: Авг. 24, 2022

The Brain Imaging Data Structure (BIDS) established community consensus on the organization of data and metadata for several neuroimaging modalities. Traditionally, BIDS had a strong focus functional magnetic resonance imaging (MRI) datasets lacked guidance how to store multimodal structural MRI datasets. Here, we present describe Extension Proposal 001 (BEP001), which adds range quantitative (qMRI) applications BIDS. In general, aim qMRI is characterize brain microstructure by quantifying physical MR parameters tissue via computational, biophysical models. By proposing this new standard, envision standardization through multicenter dissemination interoperable This way, can act as catalyst convergence between methods development application-driven studies that help develop biomarkers neural characterization. conclusion, extension offers common ground developers exchange novel tools, reducing entrance barrier in field neuroimaging.

Язык: Английский

---

Connectome-based machine learning models are vulnerable to subtle data manipulations

Patterns, Год журнала: 2023, Номер 4(7), С. 100756 - 100756

Опубликована: Май 15, 2023

Neuroimaging-based predictive models continue to improve in performance, yet a widely overlooked aspect of these is "trustworthiness," or robustness data manipulations. High trustworthiness imperative for researchers have confidence their findings and interpretations. In this work, we used functional connectomes explore how minor manipulations influence machine learning predictions. These included method falsely enhance prediction performance adversarial noise attacks designed degrade performance. Although drastically changed model the original manipulated were extremely similar (

Язык: Английский

---

Reporting checklists in neuroimaging: promoting transparency, replicability, and reproducibility

Neuropsychopharmacology, Год журнала: 2024, Номер 50(1), С. 67 - 84

Опубликована: Сен. 6, 2024

Язык: Английский

---

Precon_all: A species-agnostic automated pipeline for non-human cortical surface reconstruction

bioRxiv (Cold Spring Harbor Laboratory), Год журнала: 2025, Номер unknown

Опубликована: Апрель 20, 2025

Abstract Cortical surface reconstruction has changed how we study brain morphology and geometry. However, extending these methods to non-human species been limited by the lack of standardized pipelines, anatomical templates, variability in imaging protocols. To address challenges, present Precon_all, an open-source, species-agnostic pipeline that automates cortical for neuroimaging. It runs reliably across a wide range structures conditions successfully applied datasets from primates, carnivores, artiodactyls. Its modular framework mirrors human neuroimaging workflows, supports manual quality control, produces outputs compatible with FreeSurfer Connectome Workbench. In doing so, it substantially reduces technical barriers As data-sharing initiatives continue expand access datasets, Precon_all provides scalable solution broader adoption surface-based studying evolution through comparative neuroscience.

Язык: Английский

---

Latent diffusion for arbitrary zoom MRI super-resolution

Expert Systems with Applications, Год журнала: 2025, Номер unknown, С. 127970 - 127970

Опубликована: Май 1, 2025

---