Preliminary results with possible alternative methods (MRI)

Guylaine Collewet


This talk presents the work done on MRI acquisition, images correction,

preliminary results and future developments on image analysis. Concerning

the acquisition, the choice of all the parameters (sequence, resolution,

slice thickness, inter-slice distance, number of averages, number of images

per block) is discussed.

The method for correcting  the images in order to remove intensity

variations due to MRI imperfections is presented. Preliminary results of

image analysis using a rough segmentation algorithm based on a fixed

threshold are presented. The volume of muscle computed in the images divided

by the weight of the left carcass gives a calibration error (RMSE on all the

animals) of 1.15 % regarding the full dissection lean percentage. The limits

of this method is discussed and further improvement of the segmentation

algorithm are presented.