Scholarlink

In medical diagnostics, Single-photon emission computed tomography (SPECT) stands as an indispensable tool, yet the specter of radioactive radiation introduces potential health concerns. The quest for precise low-dose SPECT reconstruction, a linchpin in clinical applications, grapples with the formidable challenges of elevated noise and diminished spatial resolution in reconstructed images. This paper embarks on a journey to revolutionize the quality of reconstructed 2D images derived from 1D projections through the innovative lens of filtered back projection. This research’s heart beats the pulse of generative models, notably Generative Adversarial Networks (GANs) and their variations, such as Conditional Generative Adversarial Networks (CGANs). By seamlessly integrating these cutting-edge generative models, the paper sets out to elevate the reconstructed image’s quality, leveraging the prowess of a generative network as a visionary model. This transformative approach promises to metamorphose the reconstructed image, transcending conventional limitations to enhance its visual allure and overall quality. As a result, it heralds a paradigm shift in filtered back projection and paves the way for groundbreaking advancements in medical diagnostics.

错误：搜索内容不能为空，请输入英文关键词
错误：关键词超出字数限制，请精简
高级检索

Enhancing Low-Dose SPECT 2D Reconstruction from One-Dimensional Projections By Using Conditional Generative Adversarial Networks for Improved Image Quality

  • 章节

  • 发表日期：

    2025年

查看全文
  • Adis Alihodzic

摘要

In medical diagnostics, Single-photon emission computed tomography (SPECT) stands as an indispensable tool, yet the specter of radioactive radiation introduces potential health concerns. The quest for precise low-dose SPECT reconstruction, a linchpin in clinical applications, grapples with the formidable challenges of elevated noise and diminished spatial resolution in reconstructed images. This paper embarks on a journey to revolutionize the quality of reconstructed 2D images derived from 1D projections through the innovative lens of filtered back projection. This research’s heart beats the pulse of generative models, notably Generative Adversarial Networks (GANs) and their variations, such as Conditional Generative Adversarial Networks (CGANs). By seamlessly integrating these cutting-edge generative models, the paper sets out to elevate the reconstructed image’s quality, leveraging the prowess of a generative network as a visionary model. This transformative approach promises to metamorphose the reconstructed image, transcending conventional limitations to enhance its visual allure and overall quality. As a result, it heralds a paradigm shift in filtered back projection and paves the way for groundbreaking advancements in medical diagnostics.