AI based image restoration algorithm for deep tissue imaging in photoacoustic system
Photoacoustic imaging is a new medical imaging technology that is safe and radiation-free, however, there is still a need for further improvement in spatial resolution and imaging speed as photoacoustic imaging moves towards clinical application. When photoacoustic imaging systems use sparsely sampl...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis-Master by Coursework |
| Language: | English |
| Published: |
Nanyang Technological University
2023
|
| Subjects: | |
| Online Access: | https://hdl.handle.net/10356/165581 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Nanyang Technological University |
| Language: | English |
| Summary: | Photoacoustic imaging is a new medical imaging technology that is safe and radiation-free, however, there is still a need for further improvement in spatial resolution and imaging speed as photoacoustic imaging moves towards clinical application. When photoacoustic imaging systems use sparsely sampled data to reconstruct images, traditional photoacoustic image reconstruction algorithms can produce artifacts that affect the quality of the images. In this dissertation, deep learning techniques are applied to reconstruct fuzzy undersampled photoacoustic data. Using the convolutional neural network (CNN) architecture, U-net and Fully Dense U-net (FD U-net) were chosen to improve the quality of photoacoustic images. The experimental results show that both networks are capable of performing the reconstruction task and can effectively handle blurred undersampled photoacoustic microscopy images. The results produced by the two approaches are also analyzed and compared in terms of reconstructed image quality. The FD U-net, which is an improvement on the U-net, has better performance in terms of reconstruction details. |
|---|