The routine evaluation of an MR imaging for medical application is based primarily on a subjective assessment of differences in the contrast and signal intensity. Since a few years, we observe a transition from morphological imaging to quantization taking advantage of all the information content of NMR. Likewise, because of the sensitivity of the NMR relaxation parameters to food composition and structure, MRI is becoming a very promising technique to study food processing. Food processing involves in most cases internal mass and heat transfer which in turn induce temperature, concentration and structure heterogeneities within food. In particular, the quantification of the water (or whatever) concentration profile from MR images requires a careful calibration procedure. The paper aims at proposing a specific quality control procedure and evaluating the accuracy of the MRI technique in quantifying the water content concentration. The effect of the measurement time and of acquisition parameters on the accuracy of the gradient concentration has been studied. At this stage of the study, it has been focused on gradients of linear behavior and of which slope was allowed to vary. The voxel volume being constant, the solution for generating a linear gradient in proton density lies on the continuous (and linear) variation of the water concentration with the position in the object. This was operated by filling each voxel volume in the MRI slice partially with water and partially with non-protonic matter (air in the present case). The continuous variation of the water volume fraction was ensured by rotating one of the phantom plane with an angle a with reference to the plane of the MRI slice. The value of the slope in the MRI object (sGL) is obviously related to the value of the a angle. The results presented were acquired with an Magnetom Open scanner operating at 0.2 T (Siemens, Erlangen, Germany). The head coil was used as the receiver coil. The images used for the determination of the water concentration profile was a standard spin echo sequence with a fixed TE = 80ms, a fixed matrix (256x256) and a bandwidth =23 Hz/pixel. From this sequence, the effect of acquisition parameters (TR/T1 ratio, slice thickness and acquisition numbers) on the water concentration profile determination were studied. The slope expected from the a angle imposed to the phantom was compared to the experimental slope identified on the MRI data; they showed quite good agreement. Limits of such a comparison were the uncertainties on both the experimental and the MRI acquisition parameters. This procedure quality control could used to evaluate the accuracy of any MRI acquisition protocol.