The paper presents an improved version of the N3 algorithm for bias field correction in medical imaging, called N4ITK. The original N3 algorithm is widely used for correcting intensity nonuniformity in MRI data. However, it has limitations in handling certain parameters and scenarios. The authors propose modifications to the N3 algorithm, including a more robust B-spline approximation method and a modified optimization strategy, to enhance its performance. These improvements allow for better handling of high field strengths and more accurate bias field correction. The N4ITK algorithm is implemented in the Insight Toolkit (ITK) of the National Institutes of Health and is publicly available. The authors evaluated N4ITK against N3 using simulated data from the Brainweb database, hyperpolarized helium-3 lung MRI data, and postmortem hippocampus data. The results showed that N4ITK outperformed N3 in terms of bias field correction accuracy, especially under higher noise levels and smaller spline distances. The study also demonstrated that N4ITK provides more accurate and stable results compared to N3, particularly in scenarios with high field strengths and complex anatomical structures. The algorithm's multiresolution approach allows for better handling of varying frequency modulations in the bias field. Additionally, N4ITK's robust B-spline approximation method improves the algorithm's ability to handle scattered data and outliers, leading to more reliable bias field correction. The authors conclude that N4ITK is a significant improvement over the original N3 algorithm, offering better performance and reliability in correcting intensity nonuniformity in medical images. The algorithm is publicly available for use, making it accessible to the research community for further development and application.The paper presents an improved version of the N3 algorithm for bias field correction in medical imaging, called N4ITK. The original N3 algorithm is widely used for correcting intensity nonuniformity in MRI data. However, it has limitations in handling certain parameters and scenarios. The authors propose modifications to the N3 algorithm, including a more robust B-spline approximation method and a modified optimization strategy, to enhance its performance. These improvements allow for better handling of high field strengths and more accurate bias field correction. The N4ITK algorithm is implemented in the Insight Toolkit (ITK) of the National Institutes of Health and is publicly available. The authors evaluated N4ITK against N3 using simulated data from the Brainweb database, hyperpolarized helium-3 lung MRI data, and postmortem hippocampus data. The results showed that N4ITK outperformed N3 in terms of bias field correction accuracy, especially under higher noise levels and smaller spline distances. The study also demonstrated that N4ITK provides more accurate and stable results compared to N3, particularly in scenarios with high field strengths and complex anatomical structures. The algorithm's multiresolution approach allows for better handling of varying frequency modulations in the bias field. Additionally, N4ITK's robust B-spline approximation method improves the algorithm's ability to handle scattered data and outliers, leading to more reliable bias field correction. The authors conclude that N4ITK is a significant improvement over the original N3 algorithm, offering better performance and reliability in correcting intensity nonuniformity in medical images. The algorithm is publicly available for use, making it accessible to the research community for further development and application.