PET-Project

This project was created by Yuezhe Yang for the project "DeepPET: An Intelligent Brain PET Device with Low Radiation and High Imaging Quality".

Abstract

Positron Emission Tomography (PET) is one of the key imaging technologies in the field of nuclear medicine, utilizing positron emission to observe biological activity. PET plays an indispensable role in medical diagnosis and treatment, particularly in early disease detection and treatment outcome evaluation. However, PET technology currently faces several challenges, including improving image quality, reducing radiation dose, and shortening scanning time. These challenges have limited the widespread clinical application of PET. To address these issues, current research is primarily focused on hardware improvements and optimization of image reconstruction algorithms.

The vision of this project is to propose an innovative PET image reconstruction method that combines physical models with deep learning techniques to enhance imaging accuracy and efficiency while reducing radiation dose.

The advantage of this project lies in its comprehensive consideration of multiple factors, including cost, image quality, imaging speed, and radiation dose, to provide a superior PET image reconstruction solution. First, our method will integrate physical models and deep learning to achieve efficient PET image reconstruction. Compared to traditional methods, our approach requires fewer sensors and more refined equipment, thus lowering costs. Second, our method enables faster image reconstruction. Deep learning techniques have demonstrated outstanding performance in image processing, significantly accelerating reconstruction and saving time. Most importantly, our method also reduces the radiation dose. By optimizing the image reconstruction algorithm, we can reduce the required radiation dose while ensuring imaging quality, thus lowering the radiation risk for patients.

Specifically, this project will focus on two areas of research. First, we will explore efficient PET image reconstruction algorithms based on both model and data. This will involve combining iterative reconstruction algorithms, attenuation correction, and deep learning techniques to achieve high-quality PET image reconstruction. Second, we will investigate artifact-removal methods enhanced by generative models. Using conditional generative models, low-count PET images will be converted into high-count images, improving the performance of subsequent registration and producing artifact-free PET images. These studies will open up new possibilities for advancing PET technology and its applications, contributing to the development of the healthcare sector.

The primary target users of this project are equipment manufacturers and large hospitals. Our vision is to provide superior PET image reconstruction solutions that offer more accurate and reliable assistance, thereby promoting progress in the healthcare field.

There are several competitors in the PET image reconstruction space, including other research teams, medical equipment manufacturers, and research institutions. However, we believe that through our unique methods and technologies, we can maintain a competitive edge in this field. We are committed to contributing to the advancement of healthcare by improving the quality and efficiency of PET imaging, facilitating early disease detection and treatment.

Achievements

• January 2024: Our project, Learning-Based Brain PET Image Reconstruction Method and Interpretability Study successfully secured a university-level grant under the 2024 Undergraduate Innovation and Entrepreneurship Training Program!
  Project Number: X20240759 (Appliacation Link)
• April 2024: Our invention patent, PET Image Reconstruction Method Based on Prior Images and 3D Perception Method for PET Images was successfully filed!
  Patent Application Number: 202410439671.1 (Patent Link)
• July 2024: Our project, Learning-Based Brain PET Image Reconstruction Method and Interpretability Study was selected for the 2024 National Undergraduate Innovation and Entrepreneurship Training Program!
  Project Number: 202410357131(Mid-term Progress Report Link)
• July 2024: Our work, DeepPET: An Intelligent Brain PET Device with Low Radiation and High Imaging Quality, won the Gold Award in the 2024 Anhui University Innovation Competition for College Students!
  Certificate Number: HLW-2024090574
• July 2024: Our invention patent, PET Image Reconstruction Method Based on Prior Images and 3D Perception Method for PET Images was successfully published and entered substantive examination.
  Publication Serial Number: 2024073000241600
• August 2024: Our work, DeepPET: An Intelligent Brain PET Device with Low Radiation and High Imaging Quality, won the Gold Award in the Grand Final of the 2024 Anhui Provincial College Students Innovation Competition (Higher Education Main Track)!
  Certificate Number: 202410019
• September 2024: Our project, DeepPET: A Brain PET Image Reconstruction Algorithm and Device Based on Deep Learning, won the Bronze Award in the 14th "Challenge Cup" Entrepreneurship Competition for College Students at Anhui University!
  Certificate Number: 2024-TZB14-AHU049

Citation

@misc{PET-Project,
	title = {DeepPET: An Intelligent Brain PET Device with Low Radiation and High Imaging Quality},
	author = {Yuezhe Yang},
	year = {2024},
	url = {https://github.com/Bean-Young/PET-Project},
}