Denoising of low-dose chest computed tomography images using a U-net based convolutional autoencoder and transfer...

Objective.Low-dose computed tomography (LDCT) is a widely used imaging modality for performing CT examinations with reduced radiation exposure to patients. However, it suffers from increased image noise and artifacts compared to standard-dose imaging, which can hinder its diagnostic efficacy. Deep learning (DL) approaches have proven effective in LDCT denoising; however, they typically require large training datasets, which are difficult to obtain in clinical settings, particularly for denoising tasks. In this paper, a lightweight and versatile DL-based algorithm for chest LDCT denoising is proposed, trained using a two-step strategy to address the challenge of limited training data.Approach.We introduce a U-Net-based convolutional autoencoder (UNbCAE) designed to reduce noise in chest LDCT scans while preserving anatomical structures and subtle, low-contrast features. The model is first trained on phantom images and then adapted to clinical data using a transfer learning approach, thereby reducing reliance on large clinical datasets populated by paired low-dose and standard-dose images.Main results.Quantitative and qualitative experiments conducted on the LUng Nodule Analysis 2016 dataset demonstrate that the proposed method enhances image quality and improves the detectability of pulmonary nodules. Furthermore, our training strategy achieved performance comparable to or exceeding that of current state-of-the-art denoising techniques, yielding an average noise reduction factor of(3.4±0.6).Significance.The UNbCAE model provides high-quality LDCT denoising while mitigating issues related to dataset size limitation through an effective transfer learning scheme, minimizing dependence on simulated data and supporting clinical applicability.