Multiobjective evolutionary algorithms: a comparative case studyand the strength Pareto approach
IEEE Transactions on Evolutionary Computation
| Hi-index | 0.00 |
We have developed a full-scale mathematical image quality (IQ) analysis model as a tool for X-ray system design, with the view of IQ optimization and patient dose (PD) reduction. The model is implemented in LabVIEW ® with distinguishable components and processes, which allows isolation of sub-systems and exclusion of devices. All relevant PD and IQ items such as X-ray exposure, contrast, sharpness, lag and noise are calculated and additionally combined in "figures of merit" (FOM). Combining the IQ simulation model with a Pareto trade-off algorithm appears to be a promising optimization approach. Usually, multi-objective optimization studies concentrate on methods, e.g. evolutionary or genetic algorithms, with the object to reach the optimal conditions as fast and accurate as possible, preferably in a single run. However, such an approach gives less insight in the underlying principles that lead to the optimal, but also non-optimal conditions. In this study we apply the Pareto optimality with certain tolerances. Such a procedure allows optimization for conditions that are less strictly defined. This applies in particular to medical imaging, dealing with rather undefined patients as object of investigation. Further, we explore the objective domain in non-optimal regions.