N4ITK: Improved N3 Bias Correction
By Nicholas J. Tustison, Brian B. Avants, P. Cook, Yuanjie Zheng, A. Egan, Paul Yushkevich, James C. Gee. 2010
5,727 citations recorded by the source corpus.
A variant of the popular nonparametric nonuniform intensity normalization (N3) algorithm is proposed for bias field correction. Given the superb performance of N3 and its public availability, it has been the subject of several evaluation studies. These studies have demonstrated the importance of certain parameters associated with the B-spline least-squares fitting. We propose the substitution of a recently developed fast and robust B-spline approximation routine and a modified hierarchical optimization scheme for improved bias field correction over the original N3 algorithm. Similar to the N3 algorithm, we also make the source code, testing, and technical documentation of our contribution, which we denote as ¿N4ITK,¿ available to the public through the Insight Toolkit of the National Institutes of Health. Performance assessment is demonstrated using simulated data from the publicly available Brainweb database, hyperpolarized 3He lung image data, and 9.4T postmortem hippocampus data.
Subfields: Computational Mechanics, Computer Vision and Pattern Recognition, Spectroscopy
How researchers see this paper
The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.
In Computational Mechanics
- Dinggang Shen
- Davood Karimi
- Sébastien Ourselin
- Christos Davatzikos
- Marius Staring
- Stefan Klein
- Nassir Navab
- Fausto Milletarì
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dinggang Shen
- Alan L. Yuille
- Chunhua Shen
- Ming-Hsuan Yang
- Dacheng Tao
- Yi Yang
- Andrew Zisserman
In Spectroscopy
Citations in the corpus
- V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation (2016, methodology): “Datasets were first normalised using the N4 bias field correction function [21] and then resampled to a common resolution of 1 × 1 × 1.”
- Brain Tumor Segmentation Using Convolutional Neural Networks in MRI Images (2016, methodology): “To correct it, we applied the N4ITK method [6].”
- Data-Efficient Mutual Information Neural Estimator (2019, methodology): “T1-weighted images were corrected for intensity non-uniformity using N4 bias field correction [49] and skull-stripped using ANTs [3].”
- A deep learning model integrating FCNNs and CRFs for brain tumor segmentation (2017, methodology): “We pre-processed each MR scan using N4ITK and intensity normalization, which normalized each MR image’s intensity mainly by subtracting the gray-value of the highest frequency and dividing the robust deviation.”
- Scalable High Performance Image Registration Framework by Unsupervised Deep Feature Representations Learning (2015, methodology): “All images were preprocessed by the following steps: 1) manual skull removal; 2) inhomogeneity correction using N4 bias correction [64]; 3) intensity normalization for making image contrast and luminance consistent across all subjects [65]; and 4) affine registration to the selected template by FSL.”
- Objective Evaluation of Multiple Sclerosis Lesion Segmentation using a Data Management and Processing Infrastructure (2018, positive): “• Bias field correction of each modality using the N4 algorithm [39]”
- The Applications of Radiomics in Precision Diagnosis and Treatment of Oncology: Opportunities and Challenges (2019, background): “Moreover, some mature algorithms and tools could help to remove unwanted low-frequency intensity nonuniformity [41, 42] and perform image intensity normalization [43-45] in order to reduce the difference caused by different acquisition parameters of brain MRI.”
- Cross-modality image synthesis from unpaired data using CycleGAN: Effects of gradient consistency loss and training data size (2018, methodology): “N4ITK intensity inhomogeneity correction [10] was applied to all MRI volumes.”
- Statistical harmonization corrects site effects in functional connectivity measurements from multi-site fMRI data (2018, methodology): “The workflow is sketched out as follows: (1) N4 bias correction to minimize field inhomogeneity (Tustison et al., 2010); (2) brain extraction using an optimal population-specific template created by a Symmetric Group Normalization framework (Avants et al., 2010); (3) Atropos probabilistic…”
- Convolutional Neural Networks for Classification of Alzheimer's Disease: Overview and Reproducible Evaluation (2019, methodology): “The N4ITK method (Tustison et al., 2010) was used for bias field correction.”
- Computer-Aided Diagnostic System for Early Detection of Acute Renal Transplant Rejection Using Diffusion-Weighted MRI (2019, positive): “The noise and inconsistencies due to low-frequency nonuniformity, or inhomogeneity of intensities, are suppressed in part by preprocessing that combines histogram equalization with non-parametric bias correction [29].”
- Fully Automatic Brain Tumor Segmentation using End-To-End Incremental Deep Neural Networks in MRI images (2018, positive): “Moreover, the use of specific operations in the preprocessing step: N4ITK bias correction [48] as used in ([40], [37] ), and the use the post-processing step: removing flat blobs near the skull based on connected components as used in the model of [40], leads to obtaining models depend significantly”
- A review on automatic fetal and neonatal brain MRI segmentation (2017, positive): “, 1998) and N4 algorithms (Tustison et al., 2010) are commonly used to perform INU correction in the perinatal brain (Makropoulos et al.”
- Standardized evaluation of algorithms for computer-aided diagnosis of dementia based on structural MRI: the CADDementia challenge (2015, methodology): “Next to the original anonymized T1w scans, we provided these scans after non-uniformity correction with N4ITK (Tustison et al., 2010) using the following settings: shrink factor = 4, number of iterations = 150, convergence threshold = 0.00001, and initial b-spline mesh resolution = 50 mm. Images…”