Local Contrast Normalization to Improve Preprocessing in MRI of the Brain

• Published in: Bioengineering and Biomedical Signal and Image Processing pp. 255 - 266
• Main Authors: Placidi, Giuseppe, Polsinelli, Matteo
• Format: Book Chapter
• Language: English
• Published: Cham Springer International Publishing 2021
• Series: Lecture Notes in Computer Science
• Subjects: Image interpretation; Image segmentation; Local contrast normalization; Medical imaging; MRI
• ISBN: 3030881628; 9783030881627
• ISSN: 0302-9743; 1611-3349
• DOI: 10.1007/978-3-030-88163-4_23

Magnetic resonance imaging (MRI) is a fundamental medical tool for its versatility and richness of parameters. This allows the implementation of several imaging sequences capable to create high contrast images. However, contrast is also modified by magnetic field strength, system manufacturer and internal properties of the imaged body. This implies that MR images have not standardized amplitudes, though contrast normalization could help in processing and interpretation, especially when these are performed by automated strategies. We present a local contrast normalization strategy for a specific MRI imaging sequence, the FLuid Attenuated Inverse Recovery (FLAIR), one of the imaging sequence used to study inflammatory processes of the brain. The application of the proposed strategy on the images from different MRI scanners are reported and compared. Results are reported and discussed. The proposed strategy could greatly improve automatic interpretation because it reduces data variability.