Super-resolution image reconstruction with reduced computational complexity

We have recently developed a model-based adaptive image reconstruction algorithm, called dTONE, which significantly increases both image contrast and resolution by conducting a global optimization based on a model of sparse hypothetical scatterer locations. Due to the large scale and O(n³) computational complexity, images of clinically significant dimensions would require roughly 140,000 hours to reconstruct on a 3 GHz desktop computer. Here we describe an alternate algorithm - called quick TONE, or qTONE - that provides a significant reduction in computational cost with improved image contrast under the conditions tested. Both methods, along with conventional image reconstruction, were applied to tissue mimicking phantoms and excised animal tissue. For the model used, qTONE provided an 8x reduction in computation time and an 8 dB improvement in cystic contrast over the dTONE image. As the model dimensions increase, this computational advantage increases, bringing the application of this super-resolution algorithm closer to clinical feasibility.