Optimization Framework for Reconstructing Biomedical Images by Efficient Sample-based Parameterization

Abstract

An efficient computational approach for optimal reconstruction of binary-type images suitable for models in various biomedical applications is developed and validated. The methodology includes derivative-free optimization supported by a set of sample solutions with customized geometry generated synthetically. The entire framework has an easy to follow design due to a nominal number of tuning parameters which makes the approach simple for practical implementation in various settings, adjusting it to new models, and enhancing the performance. High efficiency in computational time is achieved through applying the coordinate descent method to work with individual controls in the predefined custom order. This technique is shown to outperform regular gradient-based methods with applied PCA-based control space reduction in terms of both the quality of binary images and the stability of obtained solutions when noise is present in the measurement data. The efficient performance of the complete computational framework is tested in applications to 2D inverse problems of cancer detection by the electrical impedance tomography (EIT) and demonstrated its high potential for improving the overall quality of EIT-based procedures.