At the typical spatial resolution of MRI, approximately 60-90% of voxels in the human brain contain multiple fibre populations. Quantifying microstructural properties of distinct fibre bundles within a voxel is challenging. While progress has been made for diffusion and T1-relaxation properties, resolving intra-voxel T2 heterogeneity remains an open question. Here we proposed a novel framework, COMMIT-T2, that uses tractography-based spatial regularization. Unlike previously-proposed voxel-based methods, COMMIT-T2 can recover bundle-specific T2 values within a voxel. Adding this new dimension to the microstructural characterisation of white matter pathways improves the power of tractometry to detect subtle differences in tissue properties.