MRI is the tool that is the most effectively adapted to the non-destructive analysis of fat distribution and content at the voxel scale. 

To investigate fat distribution, most MRI applications use contrasts in relaxation rates (T1 or T2) or in chemical shift (difference in NMR resonance frequency) between lipids and other tissues.    

An original approach using diffusion weighted MRI was developed to image fat distribution in muscles.  This technique is based on the considerable difference between the apparent diffusion coefficients of water and those of lipids, since water diffuses much more rapidly than lipids as a result of their different molecular weights.  With this diffusion weighted technique, the more the protons are mobile (protons in water molecules present in muscular fibre), the more their signal is reduced.  It is therefore possible to optimise the imaging method so as to turn off the water signal and, thus, to specifically image the lipids.     
The main advantage of this technique is that it makes it possible to obtain a large spatial resolution.  Its major disadvantage is its sensitivity to movements, which is generally not a problem when studying foods.

We applied this technique to imaging lipids in beef that was shallow-fried at 130 and 180°C. 

For this study, diffusion weighted MRI was conducted at 400 MHz within a microimaging system (Bruker, GmbH, Ettlingen, Germany) of the INRA platform, Magnetic Resonance Imaging of Biological Systems.  This platform includes the only broadline NMR equipment of its kind in France, allowing us to obtain previously unknown information about the structure and properties of foods, as well as to optimise their construction.   
This high-resolution imagery (0.5x0.5x0.1 mm) enabled us to determine the oil absorption profile with a much greater precision (0.1 mm) than that of biochemical analysis methods that require slicing according to the direction of penetration.

This generic method could be applied when knowledge of the distribution of lipids in a food product is necessary to optimise a process (cooking or curing) or a formulation. 

These results are therefore a major contribution to the improvement of product quality, particularly sensory and nutritional, and to the development of new foods.    



Central longitudinal section of a cylinder of raw beef (above), and beef fried at 130°C (below).  The oil on the frying face as well as the intramuscular fat is visible in hypersignal.  Each image consists of a matrix of 64 x 320 pixels corresponding to a field of vision of 32 x 32 mm.