Options
2022
Journal Article
Title
Ultralow‐parameter denoising: Trainable bilateral filter layers in computed tomography
Abstract
Background: Computed tomography (CT) is widely used as an imaging toolto visualize three-dimensional structures with expressive bone-soft tissue contrast. However, CT resolution can be severely degraded through low-doseacquisitions, highlighting the importance of effective denoising algorithms.
Purpose:Most data-driven denoising techniques are based on deep neural networks, and therefore, contain hundreds of thousands of trainableparameters, making them incomprehensible and prone to prediction failures. Developing understandable and robust denoising algorithms achieving state-of-the-art performance helps to minimize radiation dose while maintaining dataintegrity.
Methods: This work presents an open-source CT denoising framework basedon the idea of bilateral filtering. We propose a bilateral filter that can be incor-porated into any deep learning pipeline and optimized in a purely data-drivenway by calculating the gradient flow toward its hyperparameters and its input. Denoising in pure image-to-image pipelines and across different domains suchas raw detector data and reconstructed volume, using a differentiable backpro-jection layer, is demonstrated. In contrast to other models, our bilateral filter layer consists of only four trainable parameters and constrains the applied operationto follow the traditional bilateral filter algorithm by design.
Results: Although only using three spatial parameters and one intensity rangeparameter per filter layer, the proposed denoising pipelines can compete withdeep state-of-the-art denoising architectures with several hundred thousandparameters. Competitive denoising performance is achieved on x-ray microscope bone data and the 2016 Low Dose CT Grand Challenge data set. Wereport structural similarity index measures of 0.7094 and 0.9674 and peaksignal-to-noise ratio values of 33.17 and 43.07 on the respective data sets.
Conclusions: Due to the extremely low number of trainable parameters with well-defined effect, prediction reliance and data integrity is guaranteed at anytime in the proposed pipelines, in contrast to most other deep learning-based denoising architectures.
Purpose:Most data-driven denoising techniques are based on deep neural networks, and therefore, contain hundreds of thousands of trainableparameters, making them incomprehensible and prone to prediction failures. Developing understandable and robust denoising algorithms achieving state-of-the-art performance helps to minimize radiation dose while maintaining dataintegrity.
Methods: This work presents an open-source CT denoising framework basedon the idea of bilateral filtering. We propose a bilateral filter that can be incor-porated into any deep learning pipeline and optimized in a purely data-drivenway by calculating the gradient flow toward its hyperparameters and its input. Denoising in pure image-to-image pipelines and across different domains suchas raw detector data and reconstructed volume, using a differentiable backpro-jection layer, is demonstrated. In contrast to other models, our bilateral filter layer consists of only four trainable parameters and constrains the applied operationto follow the traditional bilateral filter algorithm by design.
Results: Although only using three spatial parameters and one intensity rangeparameter per filter layer, the proposed denoising pipelines can compete withdeep state-of-the-art denoising architectures with several hundred thousandparameters. Competitive denoising performance is achieved on x-ray microscope bone data and the 2016 Low Dose CT Grand Challenge data set. Wereport structural similarity index measures of 0.7094 and 0.9674 and peaksignal-to-noise ratio values of 33.17 and 43.07 on the respective data sets.
Conclusions: Due to the extremely low number of trainable parameters with well-defined effect, prediction reliance and data integrity is guaranteed at anytime in the proposed pipelines, in contrast to most other deep learning-based denoising architectures.
Author(s)