Reduction of acquisition time in intersection profile method for 4D-MRI reconstruction of thoracoabdominal organs

摘要

In this study, we developed an automatic reconstruction of 4D-MRI using the intersection profile method. This method requires several number of time sequential slices in sagittal plane and functions as a navigator slice. Data slices along the anterior and posterior axis are also captured to obtain several respiratory motions. Since the navigator slice and data slices are orthogonal to each other, it is possible to reconstruct a 4D-MRI based on the intersection of the respiratory patterns found in them. The synchronization is performed by calculating normalized cross correlation (NCC) between the respiratory pattern found in the navigator slice and data slices. In this study, we reduce the time acquisition for data slices using two approaches, namely a fixed threshold and an adaptive threshold. Fixed threshold sets a threshold value and terminate the acquisition once the NCC is value over the threshold value. Adaptive threshold determines threshold value based on the NCC value from the first 100 data slices frames. We perform both approaches to build 4D-MRI of six subjects and evaluate the geometric mean and total time to obtain all data slices compared with conventional data slices acquisition (without threshold). The preliminary results to reconstruct 4D-MRI using both approaches show a significant improvement in data slices time acquisition. Fixed threshold approach was able to reduce the time acquisition by 45% on average and 47% using adaptive threshold approach. Although the geometric mean value using both approaches were lower, the visualization of 4D-MRI using conventional data acquisition, fixed threshold and adaptive threshold showed insignificant difference.