چكيده به لاتين
The temporal super-resolution of the dynamic ultrasound imaging, a means to observe rapid heart movements is considered an important subject in medical diagnostic goals. In this study, three new techniques using the compressive sensing (CS) and matrix completion (MC) theories are introduced for the frame rate enhancement of 2-dimensional (2D) and 3-dimensional (3D) dynamic ultrasound imaging. These proposed methods are based on the acquisition scheme. In the proposed procedures, the lower scan lines for each frame are acquired. Subsequently, the CS and MC theories by the proposed methods are used to reconstruct the skipped scan lines for each frame. Two methods are proposed using the CS theory for the reconstruction of the skipped RF lines. In the first proposed method, the reconstruction of the RF lines in the RF image sequence is performed in both spatial and temporal directions. Then, the average of two reconstructions for each sample is used as the final reconstruction value. In the second proposed method, the reconstruction of the RF lines performed only using the temporal and spatial information in the neighborhood samples of a desired sample in the RF image sequences. In both proposed methods, the combination of the spatial and temporal information are used for the better reconstruction of the skipped RF lines. The qualitative and quantitative results on 2D and 3D data show that in the proposed methods, reconstruction is more accurate than the conventional spatial CS method. The quality of these videos is also suitable for medical diagnosis. According to the results, the frame rate enhancement can be improved by the factor of ten with an acceptable quality for medical diagnosis. Here, three type sparsity bases are also investigated for the sparse representation of the signals in the spatio-temporal domain, including fixed sparsity basis, fixed overcomplete dictionary and learned overcomplete dictionary. In the proposed method using the MC theory, the spatial and temporal relation of RF images and the MC theory are used for the better reconstruction of the skipped scan lines. The qualitative and quantitative evaluation of 2D and 3D data demonstrate that in the proposed method, the processing time is very low in comparison to the proposed methods using the CS theory. The implementations of all three proposed methods do not require additional of any or alteration of the hardware. In fact, the frame rate can be improved by only implementing an effective programming on the ultrasound devices. Finally, the frame rate enhancement of Duplex imaging is suggested. Our aim is the using of the CS theory for saving the time in the acquisition of the Doppler signals and therefore the frame rate enhancement of the B-mode images. Here, the proper choice of two key bases of CS including sampling and sparsity bases is explored. First, the idea of the combination of different sparsity bases is introduced for the sparsifying of the Doppler signals. Then, the idea of using the adaptive random sampling according to the structure of the Doppler signals is suggested instead of using the uniform random sampling to improve the reconstruction quality. The simulation and in-vivo data results show with the appropriate choice of sparsity and sampling bases in the CS reconstruction, the sampling rate can be reduced in the Doppler mode, and the frame rate of the Duplex imaging can be enhanced.
