Image Preprocessing for Pathological Brain Detection

TLDR: This chapter first introduces the concept of k-space, where the acquired signal lies, and reconstruction is necessary to transform it to spatial space, which can help improve the visual quality of magnetic brain images.

Abstract: Image preprocessing is quite important This chapter first introduces the concept of k-space, where the acquired signal lies First, reconstruction is necessary to transform it to spatial space Then, image denoising techniques are required Magnetic resonance images are contaminated by Rician noise in addition to common Gaussian noise Several denoising methods are introduced here A brain extraction tool is introduced to strip the skull and preserve only brain tissues The inter-class variance-based slice selection method is discussed, which aims to select one/several distinguishing slice(s) Spatial normalization is necessary, as it can transform a brain image to match a template Rigid and non-rigid normalization methods are introduced The intensity of normalization can improve image compatibility and facilitate comparability of scans with different settings Finally, image enhancement is introduced, which can help improve the visual quality of magnetic brain images Histogram equalization and contrast-limited adaptive histogram equalization methods are presented