NUCLEAR MAGNETIC RESONANCE

Author: ARISTIZABAL TORRES IVAN DARIO.
Year: 2005.
University: POLITÉCNICA DE VALENCIA [www.upv.es].
Place of defense: UNIVERSIDAD POLITÉCNICA DE VALENCIA.
Place of preparation: UNIVERSIDAD POLITÉCNICA DE VALENCIA.

Summary: Although many research results have been quite promising with Magnetic Resonance Imaging (MRI) for determining the internal quality of fruit and vegetables, there are still problems to be solved in order to achieve the commercial use of the same. Moreover, in general, have been used computers that have very high investment costs and maintenance. The RMI teams based on low-intensity magnetic field is an interesting alternative for its lower cost. This study used MRI techniques with a team low field (0.18 W), were developed and evaluated algorithms for processing digital images obtained from peaches and oranges, with the aim to automatically detect different internal damage, not more significant than with conventional methods destructive. We studied different MRI sequences, and through six quality criteria developed in this thesis, we selected those that helped obtain the best image quality in the interior of the fruit. Two sequences that spin, a weighted T1, and other weighted T2 allowed to obtain high quality images RMI at the equatorial and longitudinal both peaches and oranges, viewing them other types of internal defects. In peaches, the algorithms developed detected 98% of fruit affected by fungi aprofito, with a false detection in fruit healthy from 21%. The detection of fruit fly bite was 71%, with a high false allocation of healthy fruit with damage (42%). For the damage by cold, algorithms detected more than 98% of the fruit with damage stored for 32 days 5Â ° C, from day 22 when this happened and two days before the assessment methods destructive. In oranges, algorithms for detecting damage by fungi allowed to discriminate more than 80% of fruit affected, with a false allocation fruit as healthy damaged less than 20%. Automatic detection and early deterioration produced by fungi was conducted between 2 and 7 days in advance, before symptoms appear outside of rot. The algorithms developed for the damage by frost oranges presented in the best results of all the defects studied, with a 94% detection wise and 8% false detection, on average. Considering the results of this investigation were presented at the end a proposal to develop an automated system for inspecting the internal quality of fruit, based on MRI images achieved with a team low magnetic field.