CONTRAST MAPPING AND STATISTICAL TESTING FOR LOW-GRADE GLIOMA GROWTH QUANTIFICATION ON BRAIN MRI

A statistical differential analysis framework of longitudinal MRI volumes is proposed, based on difference maps after non-linear contrast midway mapping, to quantify tumor growth. This mapping was used to normalize MRI scans to a common range of values, and was adapted in this work to handle multiplicative MRI inhomogeneity fields. This lead to two direct applications: (1) change detection from a statistical test on differences in midway-mapped MRI data, and (2) tumoral growth quantification. A clinical evaluation was performed on 32 clinical cases with low-grade glioma, screened with two FLAIR MRI scans, several months apart. Three growth indices (volume, maximum radius and spherical radius) were measured and evaluated in terms of accuracy, comparing to manual tracing. Millimetric growth estimation precision was achieved with the proposed method for the spherical radius growth index.