Study On Back-Projection Algorithm For MIT Fast Image Reconstruction

Magnetic induction tomography (MIT) is a new medical imaging technology, whichimages the electrical conductivity distribution of the target based on impedancemeasurements. Its noninvasive, non-contact and function monitoring characteristics havereceived domestic and international research scholar’s attention. However, this technologyhas not been used clinically at present. The main technical difficulties are that highprecision measurement system and high resolution image reconstruction algorithms areneeded. Among them, the image reconstruction algorithm plays an important role inrealizing the MIT fast and accuracy imaging. An improved back-projection imagereconstruction algorithm based on the magnetic field lines distribution is presented in thispaper, which aims at solving the image reconstruction problems such as the slow speed,inaccurate positioning and the bad anti-noise performance issues. Meanwhile, the dynamicsequence images are constructed combining with the information extraction method.An imaging model with the diameter of0.2m was designed based on the MITprinciple, then the finite element method was used to solve the eddy current distribution,and the phase shifts in the half circular ring were obtained by the detection coils, and thenbe used for the image reconstruction.In order to improve the positioning accuracy of this algorithm, a data modificationmodel was built for MIT based on the electromagnetic relation. The model was used tomodify the phase shifts detected by the detection coils, which would make improvement onthe positioning accuracy. Meanwhile, the back-projection path was determined by themagnetic field line distribution in the imaging field, which would reduce the positioningerror compared with the straight line back-projection algorithm.In order to simulate the monitoring function of MIT, two groups of sequence imageswere reconstructed respectively. Combining with the conjoint analysis, the longitudinalimpedance change information is obtained, which reflects the dynamic information of theimaging target over time.The results show that the improved back-projection algorithm for MIT owns the character of high speed and strong anti-noise, which performs well in reflecting locationand shape information of the perturbations. For the sequence images, a perturbationmoving with5mm equidistance can be distinguished, this algorithm also performs wellwhen the conductivity increased with0.1S/m, which will lay a good foundation for MITclinical application.