Multi-exponential MRI T2 maps: A tool to classify and characterize fruit tissues

The estimation of multi-exponential relaxation time T2 and their associated amplitudes A0 at the voxel level has been made possible by recent developments in the field of image processing. These data are of great interest for the characterization of biological tissues, such as fruit tissues. However, they represent a high number of information, not easily interpretable. Moreover, the non-uniformity of the MRI images, which mainly directly impacts A0, could induce interpretation errors. In this paper, we propose a post-processing scheme that clusters similar voxels according to the multi-exponential relaxation parameters in order to reduce the complexity of the information while avoiding the problems associated with intensity non-uniformity. We also suggest a data representation suitable for the visualization of the multi-T2 distribution within each tissue. We illustrate this work with results for different fruits, demonstrating the great potential of multi-T2 information to shed new light on fruit characterization.

Automated summary

Recent developments in image processing have allowed for the estimation of multi-exponential relaxation time T2 and their associated amplitudes A0 at the voxel level. However, these data represent a large amount of information and are not easily interpretable, and the non-uniformity of MRI images may lead to interpretation errors. In this paper, a post-processing scheme that clusters similar voxels according to the multi-exponential relaxation parameters is proposed to reduce the complexity of the information and avoid problems associated with intensity non-uniformity. A data representation suitable for visualizing the multi-T2 distribution within each tissue is also suggested.