Deep-learning-based motion-correction algorithm in optical resolution photoacoustic microscopy

In this study, we propose a deep-learning-based method to correct motion artifacts in optical resolution photoacoustic microscopy (OR-PAM). The method is a convolutional neural network that establishes an end-to-end map from input raw data with motion artifacts to output corrected images. First, we performed simulation studies to evaluate the feasibility and effectiveness of the proposed method. Second, we employed this method to process images of rat brain vessels with multiple motion artifacts to evaluate its performance for in vivo applications. The results demonstrate that this method works well for both large blood vessels and capillary networks. In comparison with traditional methods, the proposed method in this study can be easily modified to satisfy different scenarios of motion corrections in OR-PAM by revising the training sets.

The celestial reference frame stability and apparent motions of the radio sources

Time series of the coordinates of the ICRF radio sources were analyzed. It was shown that part of radio sources, including even the so-called “defining” sources, show a the significant apparent motion. Corrections for their a priori coordinates are time functions. The celestial reference frame stability is provided by the no-net-rotation condition applied to the selected subset of sources, which leads in our case to a rotation of the frame axes with time. Parameters of this rotation were calculated for different subsets of sources.To improve stability of the celestial reference frame new methods of selection of the extragalactic radio sources were suggested. The first one was called “cosmological” and the second one “kinematical”. It was shown that a selected subset of the ICRF sources, according to cosmological criteria, determines the most stable coordinate system during next decade.

The Characteristics of Key Analysis Errors. Part II: The Importance of the PV Corrections and the Impact of Balance

This study examines a few approaches to isolate the balanced component of the initial corrections from the Canadian Meteorological Centre energy-norm-based key analysis error algorithm, in an attempt to capture the part of the key analysis errors responsible for short-range forecast errors. The best results were obtained with the nonlinear balance potential vorticity (PV) inversion technique. It was shown that the PV component of the initial corrections contains the essential information for reducing short-range forecast errors. The remaining imbalance part of the initial corrections does not grow in time and does not contribute to the improvement of the forecast. The removal of the imbalance part of the initial corrections makes the corrected analysis slightly closer to the observations, but remains systematically farther away as compared with the original analysis. Thus the balanced part of the key analysis errors cannot justifiably be associated to analysis errors. A methodology to balance the divergent part of the initial corrections, which reduces significantly the spinup in the vertical motion corrections, is also presented. Finally, in light of the results presented in this paper, some recommendations to improve the key analysis error algorithm are proposed.