正子造影影像重建與軟體介面開發

Abstract

在本論文中，開發了 ASPET 軟體來增強 PET 資料分析與視覺化，滿足真實資料及模擬影像的需求。該軟體具有簡潔的介面，允許應用各種重建演算法，從傳統的濾波反投影 (FBP) 到最大似然期望最大化 (MLEM) 以及三角度 MLEM 等進階迭代方法。迭代演算法如 MLEM 與三角度 MLEM，相較於 FBP，其影像品質有顯著改進，有效減少了雜訊與偽影，同時提升成像解析度。本研究凸顯 ASPET 軟體顯著改善 PET 影像的潛力，有助益於醫學領域的科學家與臨床醫生。ASPET 軟體採用模組化設計，確保軟體高度延展性，可整合新演算法並提高大型資料集和複雜迭代過程的運算效率。重建結果方面，展現了根據特定成像要求，選擇適當重建演算法的重要性。雖然 FBP 速度快且簡單，但在處理雜訊及生成高解析度影像方面存在局限性，不太適用於臨床診斷。相比之下，MLEM 與三角度 MLEM 提供更準確、高品質的影像，更適合臨床應用。ASPET 軟體的持續開發與維護，旨在為科學研究與臨床診斷提供良好的使用體驗，軟體功能強化可促進更精確、更可靠的 PET 影像重建品質，推動醫學影像及相關領域的創新。

In this thesis, the ASPET software has been developed to enhance the analysis and visualization of PET data, catering to both real-world and simulation scenarios. The software features a user-friendly interface, allowing users to apply various reconstruction algorithms, from conventional Filtered Back Projection (FBP) to advanced iterative methods like Maximum Likelihood Expectation Maximization (MLEM) and three-angle MLEM. The iterative algorithms, particularly MLEM and three-angle MLEM, demonstrate significant improvements in image quality over FBP, effectively reducing noise and artifacts while enhancing resolution. This work highlights the ASPET software's potential to significantly improve PET imaging, offering essential tools for scientists and clinicians in the medical field. The modular design of ASPET software ensures adaptability, allowing integration of new algorithms and enhancing computational efficiency for large datasets and complex iterative processes. The results of this study underscore the importance of selecting appropriate reconstruction algorithms based on specific imaging requirements. While FBP offers speed and simplicity, it is less suitable for detailed clinical diagnostics due to its limitations in handling noise and producing high-resolution images. In contrast, MLEM and three-angle MLEM provide more accurate and high quality images, making them preferable for clinical applications. This continuous development aims to keep ASPET at the cutting edge of PET imaging technology, offering robust support for clinical diagnostics and treatment planning. Additionally, the software's enhancements will contribute to scientific research, facilitating more precise and reliable PET imaging and driving innovations in medical imaging and related fields.