In this work, we combine a pure phase-encoded Magnetic Resonance Imagi
ng (MRI) method with a new tissue-classification technique to make geo
metric models of a human tooth, We demonstrate the feasibility of thre
e-dimensional imaging of solids using a conventional 11.7-T NMR spectr
ometer. In solid-state imaging, confounding line-broadening effects ar
e typically eliminated using coherent averaging methods. Instead, we c
ircumvent them by detecting the proton signal at a fixed phase-encode
time following the radio-frequency excitation. By a judicious choice o
f the phase-encode time in the MR imaging protocol, we differentiate e
namel and dentine sufficiently to successfully apply a new classificat
ion algorithm. This: tissue-classification algorithm identifies the di
stribution of different material types, such as enamel and dentine, in
volumetric data. In this algorithm, we treat a voxel as a volume, not
as a single point, and assume that each voxel may contain more than o
ne material. We use the distribution of MR image intensities within ea
ch voxel-sized volume to estimate the relative proportion of each mate
rial using a probabilistic approach. This combined approach, involving
MRT and data classification, is directly applicable to bone imaging a
nd hard tissue contrast-based modeling of biological solids.