A novel denoising method for low-dose CT images based on transformer and CNN

by Selleckchem | Oct 28 2023

Computed Tomography (CT) has become a mainstream imaging tool in medical diagnosis. However, the issue of increased cancer risk due to radiation exposure has raised public concern. Low-dose computed tomography (LDCT) technique is a CT scan with lower radiation dose than conventional scans. LDCT is used to make a diagnosis of lesions with the smallest dose of x-rays, and is currently mainly used for early lung cancer screening. However, LDCT has severe image noise, and these noises affect adversely the quality of medical images and thus the diagnosis of lesions. In this paper, we propose a novel LDCT image denoising method based on transformer combined with convolutional neural network (CNN). The encoder part of the network is based on CNN, which is mainly used to extract the image detail information. In the decoder part, we propose a dual-path transformer block (DPTB), which extracts the features of input of the skip connection and the features of input of the previous level through two paths respectively. DPTB can better restore the detail and structure information of the denoised image. In order to pay more attention to the key regions of the feature images extracted at the shallow level of the network, we also propose a multi-feature spatial attention block (MSAB) in the skip connection part. Experimental studies are conducted, and comparisons with the state-of-the-art networks are made, and the results demonstrate that the developed method can effectively remove the noise in CT images and improve the image quality in the evaluation metrics of peak signal to noise ratio (PSNR), structural similarity (SSIM), and root mean square error (RMSE) and is superior to the state-of-the-art models. Our method achieved 28.9720 of PSNR, 0.8595 of SSIM and 14.8657 of RMSE on the Mayo Clinic LDCT Grand Challenge dataset. For different noise level σ (15, 35, and 55) on the QIN_LUNG_CT dataset, our proposed also achieved better performances.

Comments:

The paper proposes a novel method for denoising low-dose computed tomography (LDCT) images using a combination of transformer and convolutional neural network (CNN) architectures. LDCT is a technique that reduces the radiation dose in CT scans, but it often results in severe image noise, which can affect the quality of medical images and the accuracy of lesion diagnosis.

The proposed method utilizes a CNN-based encoder to extract detailed information from the LDCT images. In the decoder part, a dual-path transformer block (DPTB) is introduced. The DPTB extracts features from the input of the skip connection and the input of the previous level through two separate paths, enabling the restoration of detailed and structural information in the denoised image.

To prioritize key regions in the feature images extracted at shallow levels of the network, a multi-feature spatial attention block (MSAB) is incorporated in the skip connection part. This block helps to focus more on important regions during the denoising process.

The paper presents experimental studies and compares the proposed method with state-of-the-art networks. The results demonstrate that the developed method effectively removes noise in CT images and improves image quality, as measured by peak signal-to-noise ratio (PSNR), structural similarity (SSIM), and root mean square error (RMSE) evaluation metrics. On the Mayo Clinic LDCT Grand Challenge dataset, the proposed method achieves a PSNR of 28.9720, SSIM of 0.8595, and RMSE of 14.8657. Furthermore, the proposed method also performs well on the QIN_LUNG_CT dataset for different noise levels (σ) of 15, 35, and 55.

Overall, the paper introduces a promising approach for LDCT image denoising, combining transformer and CNN architectures. The proposed method outperforms existing models and demonstrates its potential to enhance the quality of LDCT images, aiding in more accurate diagnosis of lesions.