Loading...

Improving the Quality and Speed Ultrasonic Image Reconstruction Using the K_SVD Algorithm

Alimoradi, Zahra | 2021

393 Viewed
  1. Type of Document: M.Sc. Thesis
  2. Language: Farsi
  3. Document No: 54235 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Jahed, Mehran; Kaveh Vash, Zahra
  7. Abstract:
  8. Ultrasound imaging is one of the most widely used diagnostic methods in medicine due to its safety, cost-effectiveness and small and portable size of equipment. In this imaging method, line-by-line scanning is used to create the image, which has limitations in speed and resolution. To deal with these limitations, new methods have been proposed, such as plane-wave imaging with the possibility of increasing the imaging speed and synthetic apertures with the possibility of increasing the quality of images. By combining these two methods, high quality and speed can be achieved simultaneously. Therefore, in this research, based on previous studies, a number of elements were activated each time, and a plane wave was sent and the response was received by all the elements. Since any random selection of array elements is a linear combination of each element, the data collected after each radiation is a linear combination of synthetic aperture method data. Furthermore, as the number of radiating elements is less than the total number of elements, using the compressed sensing structure, complete data were reconstructed by the synthetic aperture method. Although in the compressed sensing structure, the high-resolution data of the synthetic aperture is received faster, the quality of the reconstructed images is reduced compared to a reconstruction with full data. In this study, by using the K-SVD algorithm and dictionaries with learning capabilities, the resolution of reconstructed images was increased in addition to the improved speed. The results show that using the proposed method, RMSE can be reduced by 0.2. Previous studies show that the super-resolution structured illumination method can be used to pass the resolution beyond the diffraction limit, and as a result, the high-frequency data of the resulting image can lead to improved resolution. In this study, using the deconvolution method, the desired sinusoidal pattern in the super-resolution structured illumination method was created at the desired depth, which is 5 times faster than the method used in previous studies, and in addition to obtaining high-frequency information, the image reconstruction speed also increased. In this research, Field II has been used for simulations, which is a suitable tool for probe simulation and ultrasound imaging and is implemented in MATLAB environment
  9. Keywords:
  10. Synthetic Aperture ; Super Resolution ; Compressive Sensing ; Deconvolution ; Medical Ultrasound Imaging ; Plane Wave Imaging ; K-SVD Algorithm ; Learning Dictionaries ; Synthetic Aperture Imaging

 Digital Object List

 Bookmark

No TOC