DeepNeuro: an open-source deep learning toolbox for neuroimaging

Beers, Andrew; Brown, James M.; Chang, Ken; Hoebel, Katharina; Patel, Jay; Ly, K. Ina; Tolaney, Sara M.; Brastianos, Priscilla K.; Rosen, Bruce R.; Gerstner, Elizabeth R.; Kalpathy–Cramer, Jayashree

doi:10.1007/s12021-020-09477-5

Cited by 37 publications

(27 citation statements)

References 37 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…In this context, several approaches based on ANN have been proposed to improve the accuracy of brain extraction. However, these ANN algorithms have focused on learning brain extraction from training datasets either containing a collection of normal (or apparently normal) brain MRI from public datasets (Dey & Hong, ; Sadegh Mohseni Salehi, Erdogmus, & Gholipour, ), or from a limited number of (single institutional) brain MRI with pathologies (Beers et al, ; Kleesiek et al, ). Therefore, generalizability of these ANN algorithms to complex multicenter datasets may be limited on unseen data with varying MR hardware and acquisition parameters, pathologies or treatment‐induced tissue alterations.…”

Section: Introductionmentioning

confidence: 99%

Automated brain extraction of multisequence MRI using artificial neural networks

Isensee

Schell

Pflueger

et al. 2019

Human Brain Mapping

345

253

View full text Add to dashboard Cite

Brain extraction is a critical preprocessing step in the analysis of neuroimaging studies conducted with magnetic resonance imaging (MRI) and influences the accuracy of downstream analyses. The majority of brain extraction algorithms are, however, optimized for processing healthy brains and thus frequently fail in the presence of pathologically altered brain or when applied to heterogeneous MRI datasets. Here we introduce a new, rigorously validated algorithm (termed HD‐BET) relying on artificial neural networks that aim to overcome these limitations. We demonstrate that HD‐BET outperforms six popular, publicly available brain extraction algorithms in several large‐scale neuroimaging datasets, including one from a prospective multicentric trial in neuro‐oncology, yielding state‐of‐the‐art performance with median improvements of +1.16 to +2.50 points for the Dice coefficient and −0.66 to −2.51 mm for the Hausdorff distance. Importantly, the HD‐BET algorithm, which shows robust performance in the presence of pathology or treatment‐induced tissue alterations, is applicable to a broad range of MRI sequence types and is not influenced by variations in MRI hardware and acquisition parameters encountered in both research and clinical practice. For broader accessibility, the HD‐BET prediction algorithm is made freely available (http://www.neuroAI-HD.org) and may become an essential component for robust, automated, high‐throughput processing of MRI neuroimaging data.

show abstract

Section: Introductionmentioning

confidence: 99%

Automated brain extraction of multisequence MRI using artificial neural networks

Isensee

Schell

Pflueger

et al. 2019

Human Brain Mapping

345

253

View full text Add to dashboard Cite

show abstract

“…Finally, Beers et al [23] introduced a deep learning framework that puts the best fit deep learning algorithms in order to deal with neuroimaging practically. The authors noted that many packages are very good at sharing and designing deep learning algorithms, but few provide utilities for working with clinical data.…”

Section: D Skull-stripping Methodsmentioning

confidence: 99%

“…Leung et al [ [130] Deep MRI brain extraction https://github.com/GUR9000/Deep_MRI_brain_extraction Huo et al [131] SLANT https://github.com/MASILab/SLANTbrainSeg Isensee et al [132] HD-BET https://github.com/MIC-DKFZ/HD-BET Fedorov et al [133] MeshNet https://github.com/Entodi/MeshNet Beers et al [23] DeepNeuro https://github.com/QTIM-Lab/DeepNeuro Lutkenhoff et al [107] optiBET http://montilab.psych.ucla.edu/fmri-wiki…”

Section: Atlas or Library-based Methodsmentioning

confidence: 99%

“…In other words, brain images from two or more-time intervals from the same subject are compared to assess how the brain has altered over time [20,21].Manual skull stripping is a laborious task due to low contrast images, obscure boundaries of the brain in the MRI, and the absence of intensity standardization [22]. Moreover, the whole brain extraction becomes more challenging and limited in the presence of varying acquisition parameters or when a brain MRI dataset presents a pathological disorder, such as a brain tumor [23].Appl. Sci.…”

mentioning

confidence: 99%

See 1 more Smart Citation

Conventional and Deep Learning Methods for Skull Stripping in Brain MRI

2020

View full text Add to dashboard Cite

Skull stripping in brain magnetic resonance volume has recently been attracting attention due to an increased demand to develop an efficient, accurate, and general algorithm for diverse datasets of the brain. Accurate skull stripping is a critical step for neuroimaging diagnostic systems because neither the inclusion of non-brain tissues nor removal of brain parts can be corrected in subsequent steps, which results in unfixed error through subsequent analysis. The objective of this review article is to give a comprehensive overview of skull stripping approaches, including recent deep learning-based approaches. In this paper, the current methods of skull stripping have been divided into two distinct groups—conventional or classical approaches, and convolutional neural networks or deep learning approaches. The potentials of several methods are emphasized because they can be applied to standard clinical imaging protocols. Finally, current trends and future developments are addressed giving special attention to recent deep learning algorithms.

show abstract

“…Although Atlas is solving a unique problem, there are several libraries that share similarities. For example, NiftyNet [6], Nuts-ML [9], and DeepNeuro [1] were all developed to handle the preprocessing of diagnostic images exclusively for use in deep learning applications. Similarly, the Deep Learning Toolkit [10] was developed specifically for prototyping deep learning models and modules.…”

Section: Introductionmentioning

confidence: 99%

Creational and Structural Patterns in a Flexible Machine Learning Framework for Medical Ultrasound Diagnostics

Thibeault

EPiC Series in Computing

View full text Add to dashboard Cite

The impact of machine learning in medicine has arguably lagged behind its commercial counterparts. This may be attributable to the generally slower pace and higher costs associated with clinical applications, but also present are the conflicting constraints and requirements of learning from data in a highly regulated industry that introduce levels of complexity unique to the medical space. Because of this, the balance between innovation and controlled development is challenging. Adding to this are the multiple modalities found in most clinical applications where applying traditional machine learning preprocessing and cross-validation techniques can be precarious. This work presents the novel use of creational and structural design patterns in a generalized software framework intended to alleviate some of those difficulties. Designed to be a configurable pipeline to not only support the experimentation and development of diagnostic machine learning algorithms, but also to support the transition of those algorithms into production level systems in a composed manner. The resulting framework provides the foundation for developing unique tools by both novice and expert data scientists.

show abstract

DeepNeuro: an open-source deep learning toolbox for neuroimaging

Cited by 37 publications

References 37 publications

Automated brain extraction of multisequence MRI using artificial neural networks

Automated brain extraction of multisequence MRI using artificial neural networks

Conventional and Deep Learning Methods for Skull Stripping in Brain MRI

Creational and Structural Patterns in a Flexible Machine Learning Framework for Medical Ultrasound Diagnostics

Contact Info

Product

Resources

About