Methods and algorithms for reconstruction of difference images using prior structural information

Unmet Need
The global market for medical imaging should grow from $34.1 billion in 2018 to reach about $43.6 billion by 2023. One of the most important aspects of medical imaging is sequential imaging. Currently, most sequential studies have a primary goal of estimating the difference between a prior image and the current measurements of the anatomy, for example predicting the change between a scan from 3 months ago to what the organ may look like today. Sequential studies utilize prior-image-based reconstruction (PBIR) algorithms which usually incorporate the prior image information through a penalty term in the objective function. The standard penalized likelihood (PL) algorithm invokes limitations upon the function, making sure the difference contains the same constraints as the prior image. However, many of the current sequential medical imaging techniques are not very reliable because much of the data received in prior images can be noisy and sparsely sampled. In addition to this, because of the need to update the image sequentially and every few months, the dose of radiation exposure to the patient can be very high. A new method is needed to improve the reliability of the reconstructed difference as well as reduce the dosage of radiation as much as possible. 
 
Technology Overview
Johns Hopkins inventors introduced a new method called reconstruction of difference (RoD) to reconstruct the difference image directly from the measurements. The method relies on prior image data, but rather than invoking a penalty term, the information is integrated into the data consistency term. The RoD aims to target both a reduction in computation speed as well as reduction in radiation dose through its use with local acquisition.The inventors have tested the performance of RoD under noisy and sparsely sampled projections in local and global acquisition scenarios. The RoD outperformed standard PL in noisy and truncated data as well as in simulation and test-bench data reconstructions.

Stage of Development
The inventors have created a prototype of the algorithm and are working on further tests.

Publications
Pourmoteza A, et al. Physics in Medicine & Biology 61, 5, 2016
 

Category(s):