Text controllable PET denoising

TL;DR

This paper introduces a novel text-guided PET image denoising method that leverages pretrained CLIP features and a U-Net architecture to improve image quality across various noise levels, potentially reducing scan time.

Contribution

The study presents a new text-guided denoising model for PET images that works across multiple count levels within a single framework, combining CLIP and U-Net techniques.

Findings

Significant qualitative and quantitative improvements in PET image quality.

Model effectively handles a wide range of noise levels.

Potential to reduce acquisition time in PET imaging.

Abstract

Positron Emission Tomography (PET) imaging is a vital tool in medical diagnostics, offering detailed insights into molecular processes within the human body. However, PET images often suffer from complicated noise, which can obscure critical diagnostic information. The quality of the PET image is impacted by various factors including scanner hardware, image reconstruction, tracer properties, dose/count level, and acquisition time. In this study, we propose a novel text-guided denoising method capable of enhancing PET images across a wide range of count levels within a single model. The model utilized the features from a pretrained CLIP model with a U-Net based denoising model. Experimental results demonstrate that the proposed model leads significant improvements in both qualitative and quantitative assessments. The flexibility of the model shows the potential for helping more complicated denoising demands or reducing the acquisition time.