Intensity inhomogeneity correction in brain MRI: a systematic review of techniques, current trends and future challenges

Intensity inhomogeneity, a common artefact in brain magnetic resonance imaging, poses challenges in medical image analysis. Intensity inhomogeneity, also known as bias field, occurs in magnetic resonance images due to factors such as magnetic field non-uniformity, radiofrequency coil sensitivity, tissue properties, patient-related factors, scanner artefacts. It creates intensity non-uniformity inside the homogeneous tissue regions of the brain images. Thereby degrading the performance of diagnosis assessment. This systematic review proposes a first hand categorization of a range of methodologies for intensity inhomogeneity correction. In particular, an overview of retrospective techniques including the filtering methods, computational intelligence methods, fuzzy models, learning-based approaches, etc. is included. This paper also presents the emergence of learning-based techniques in developing the intensity inhomogeneity correction techniques. Additionally, major challenges, current trends, and future directions for research and development are discussed. Moreover, the characteristics of choosing a suitable method and the appropriate evaluation metric are elaborately presented. This paper may serve as a comprehensive resource for researchers, clinicians, and engineers interested in enhancing the quality and reliability of brain image analysis through effective intensity inhomogeneity correction techniques.