A performance comparison of convolutional neural network‐based image denoising methods: The effect of loss functions on low‐dose CT images

Abstract

Purpose Convolutional neural network (CNN)‐based image denoising techniques have shown promising results in low‐dose CT denoising. However, CNN often introduces blurring in denoised images when trained with a widely used pixel‐level loss function. Perceptual loss and adversarial loss have been proposed recently to further improve the image denoising performance. In this paper, we investigate the effect of different loss functions on image denoising performance using task‐based image quality assessment methods for various signals and dose levels. Methods We used a modified version of U‐net that was effective at reducing the correlated noise in CT images. The loss functions used for comparison were two pixel‐level losses (i.e., the mean‐squared error and the mean absolute error), Visual Geometry Group network‐based perceptual loss (VGG loss), adversarial loss used to train the Wasserstein generative adversarial network with gradient penalty (WGAN‐GP), and their weighted summation. Each image denoising method was applied to reconstructed images and sinogram images independently and validated using the extended cardiac‐torso (XCAT) simulation and Mayo Clinic datasets. In the XCAT simulation, we generated fan‐beam CT datasets with four different dose levels (25%, 50%, 75%, and 100% of a normal‐dose level) using 10 XCAT phantoms and inserted signals in a test set. The signals had two different shapes (spherical and spiculated), sizes (4 and 12 mm), and contrast levels (60 and 160 HU). To evaluate signal detectability, we used a detection task SNR (tSNR) calculated from a non‐prewhitening model observer with an eye filter. We also measured the noise power spectrum (NPS) and modulation transfer function (MTF) to compare the noise and signal transfer properties. Results Compared to CNNs without VGG loss, VGG‐loss‐based CNNs achieved a more similar tSNR to that of the normal‐dose CT for all signals at different dose levels except for a small signal at the 25% dose level. For a low‐contrast signal at 25% or 50% dose, adding other losses to the VGG loss showed more improved performance than only using VGG loss. The NPS shapes from VGG‐loss‐based CNN closely matched that of normal‐dose CT images while CNN without VGG loss overly reduced the mid‐high‐frequency noise power at all dose levels. MTF also showed VGG‐loss‐based CNN with better‐preserved high resolution for all dose and contrast levels. It is also observed that additional WGAN‐GP loss helps improve the noise and signal transfer properties of VGG‐loss‐based CNN. Conclusions The evaluation results using tSNR, NPS, and MTF indicate that VGG‐loss‐based CNNs are more effective than those without VGG loss for natural denoising of low‐dose images and WGAN‐GP loss improves the denoising performance of VGG‐loss‐based CNNs, which corresponds with the qualitative evaluation.

Ultra-low-dose chest CT imaging of COVID-19 patients using a deep residual neural network

“…Yang et al [ 22 ] applied a generative adversarial network (GAN) with the Wasserstein distance and perceptual loss to denoise low-dose CT images. In another study, Kim et al [ 23 ] investigated the effect of different loss functions on convolutional neural network (CNN)–based image denoising performance using task-based image quality assessment for various signals and dose levels. Shin et al [ 24 ] compared the image quality of low-dose CT images obtained using a deep learning–based denoising algorithm with low-dose CT images reconstructed using filtered backprojection (FBP) and advanced modeled iterative reconstruction (ADMIRE).…”

Low‐dose CT image and projection dataset

“…At the simulated reduced dose levels included in this library, we believe that any differences that may exist between simulated and measured data have negligible impact on algorithms developed using the provided lower‐dose data. This belief is supported by the successful use of the data in numerous publications 26–34 …”

“…A multiresolution deep learning U‐net for sparse‐view CT 30 Performance comparison of CNN‐based image denoising methods using different loss functions 31 A self‐attention CNN for low‐dose CT denoising with self‐supervised perceptual loss network 32 …”

“…Many noise reduction techniques trained based on these paired datasets have demonstrated great success in terms of reducing image noise and improving image quality. [26][27][28][29][30][31][32][33][34] However, given the recent success of unsupervised low-dose CT reconstruction, it is desirable also to have unpaired low-dose and full-dose dataset available, which is a topic of interest for future development. Meanwhile, the current paired dataset provided in this work can still be used to meet that purpose using appropriate arrangement of training cases (e.g., organize the cases in an unpaired fashion), which is beneficial as the full-dose images can be used as references even if they are not used in the training and testing.…”

Eigenloss: Combined PCA-Based Loss Function for Polyp Segmentation

“…Therefore, loss functions, also known as objective functions, cost functions, error functions or simply losses, conduct the training process of deep learning models computing the error between predicted values and ground truth and minimizing it to optimize the weights of the model. The relevance and influence of loss functions in deep learning models have been studied in many fields, such as handwritten digits recognition [47], fast single classification [48], cardiac MRI reconstruction [49], low-dose CT images denoising [50], or face recognition [51]. Loss functions can be categorized into distribution, region or boundary-based losses, depending on the main concept considered for minimization.…”