Loading...

Improving Resolution and Reducing Artifacts in Ultrasound and Optics Imaging for Computed Tomography Structures

Hakakzadeh, Soheil | 2024

0 Viewed
  1. Type of Document: Ph.D. Dissertation
  2. Language: Farsi
  3. Document No: 57591 (05)
  4. University: Sharif University of Technology
  5. Department: Electrical Engineering
  6. Advisor(s): Kavehvash, Zahra
  7. Abstract:
  8. The hybrid imaging technique combining ultrasound and optical tomography, based on computed tomography, has gained significant attention in the past two decades due to its non-invasive and non-ionizing advantages in medical imaging. By integrating optical and ultrasound imaging, it enables high-resolution imaging and diagnosis based on optical absorption properties of various tissues. However, challenges such as large transducer size, limited bandwidth, and unwanted artifacts in the imaging process limit the quality of reconstructed images and prevent achieving optimal resolution and contrast. This study focuses on developing processing and structural methods to overcome these challenges and improve image quality in ring-based photoacoustic tomography systems. To address the impact of large transducer size, which reduces tangential resolution, a Multi-Angle PAT (MA-PAT) detection structure is proposed. Theoretical analysis shows that as the distance from the center increases, the range of spatial frequency components decreases due to transducer size limitations, leading to reduced tangential resolution. In the MA-PAT approach, each transducer rotates around the scan center and at specific central angles. These rotation angles, calculated based on the transducer’s central frequency, length, and imaging region radius, help improve tangential resolution. For instance, in imaging an ROI with a 55-mm radius using a 13-mm diameter transducer with a central frequency of 2.25 MHz, the optimal number of detection points was estimated at 150. Additionally, each detection point in MA-PAT requires central rotations at ±8 and ±16 degrees. Experimental and simulation results, by comparing PSF charts at various distances from the center, showed that this approach significantly creates uniform tangential resolution across the imaging area and enhances image quality throughout the entire ROI. Focusing on the limitations posed by transducer bandwidth, which reduces axial resolution, a signal recovery algorithm was designed for photoacoustic signals. The limited bandwidth of transducers causes a reduction in image resolution and contrast, as well as the presence of side lobes and unwanted artifacts in reconstructed images. The proposed algorithm, by designing a mask for extracting signals and removing unwanted waves, enables improved axial resolution and contrast in reconstructed images. Evaluation of this method shows that using recovered signals in standard reconstruction algorithms like DAS and DMAS improves axial resolution by up to 45% and increases contrast by 1.16 dB. Additionally, the proposed method successfully reduces background artifacts by 80%, as evidenced in numerical studies, experimental tests, and in vivo imaging (including human forearm and tungsten wires). To address unwanted artifacts, two novel methods are proposed. Both methods aim to improve image quality in full-view ring-based photoacoustic imaging, overcoming limitations in detector numbers and the high cost of C-PAI systems. The first method introduces a new spatial amplitude factor to reduce streak artifacts and enhance image resolution and contrast in sparse sampling conditions. Numerical and experimental tests evaluated this method compared to conventional methods like UBP and CTBP. Using 1200 spatial samples, the SSIM index and gCNR ratio were measured at 0.97 and 0.87, respectively, showing a significant improvement over other methods. The CNR also reached 6.61 dB, marking a substantial enhancement in image quality. Even when reducing spatial samples to 240, the method maintained optimal performance, demonstrating that with a slight increase in computational load, image quality can be effectively improved. However, this method may consider regions with low photoacoustic signal amplitude (around -40 dB) as artifacts, resulting in their elimination. To address this issue, an ASM-based approach was proposed. In this method, multiple low-resolution images are generated using detection points at random locations, and with ASM, the exact artifact locations are identified and removed in the final image. This method was tested in various studies (including human vessels and brain), experimental tests (such as complex triangles and leaves), and in vivo studies (rat brain imaging). It demonstrated SNR and SSIM improvements of 20 dB and 25%, respectively, compared to conventional methods, with a 50-fold reduction in artifact standard deviation. All chapters of this research ultimately aim at a common goal: developing a high-quality photoacoustic imaging system capable of providing accurate images of complex structures. The methods presented in this study effectively enhance resolution, reduce artifacts, and improve image quality in ring-based photoacoustic imaging. Together, they contribute to the development of more cost-effective and efficient imaging systems
  9. Keywords:
  10. Photoacoustic Tomography ; Computed Tomography (CT) ; Artifacts Reduction ; Resolution Enhancement ; Image Enhancement ; Medical Ultrasound Imaging

 Digital Object List

 Bookmark

No TOC