Microstructure imaging is hard, because sensitivity of MRI measurements to interesting features of tissue microstructure, such as axon diameter or the size of cancer cells, is low, particularly in measurements from clinical MRI scanners. This project uses advanced statistical techniques to pool information from similar image voxels and thus increase sensitivity to the subtle changes we look for in microstructure imaging. Early stages of the project use random effects modelling for increasing the accuracy of estimates of crossing fibre estimates. Later stages will use the same approach for mapping axon diameter and density in live human subjects.