DLA based compressed sensing for high resolution MR microscopy of neuronal tissue

Abstract

In this work we present the implementation of compressed sensing (CS) on a high field preclinical scanner (17.2 T) using an undersampling trajectory based on the diffusion limited aggregation (DLA) random growth model. When applied to a library of images this approach performs better than the traditional undersampling based on the polynomial probability density function. In addition, we show that the method is applicable to imaging live neuronal tissues, allowing significantly shorter acquisition times while maintaining the image quality necessary for identifying the majority of neurons via an automatic cell segmentation algorithm.

Keywords: Cell segmentation; Compressed sensing (CS); Diffusion limited aggregation (DLA); Magnetic resonance imaging (MRI); Magnetic resonance microscopy (MRM); Total variation (TV).