Identification of the Uncertainty Structure to Estimate the Acoustic Release of Chemotherapeutics From Polymeric Micelles.

This paper estimates the acoustic drug release from micelles after accurately identifying the underlying statistical noise characteristics in experimental data. The drug release is measured as a change in fluorescence as ultrasound is applied. First, the noise structure affecting the process dynamics and the measurement process is identified in terms of statistical covariance of the aforementioned quantities. Then, the identified covariance magnitudes are utilized to estimate the dynamics of drug release. The performance of different filters is investigated. The identified a priori knowledge is used to implement an optimal Kalman filter, a multi-hypothesis Kalman filter, and a variant of the full information estimator (moving horizon estimator) to the problem at hand. The proposed algorithms are initially deployed in a simulation environment, and then the experimental data sets are fed into the algorithms to validate their performance. Experiments span a number of ultrasonic power densities for both non-targeted and targeted polymeric micelles (the targeting being accomplished using the folate moiety). The results suggest that the proposed algorithm, the optimal Kalman filter, performs better than the other two in all tests performed.