Assessing the consistency of CT-based ventilation imaging under noise reduction processing with simulated quantum noise using a nonrigid alveoli phantom

TL;DR

This study shows that CNN-based noise reduction improves the accuracy of CT-based ventilation imaging, especially in high-noise conditions.

Contribution

The study demonstrates that CNN-based denoising outperforms traditional methods in preserving ventilation image accuracy under quantum noise.

Findings

CNN-based denoising significantly improved categorical consistency of CTVI under high-noise conditions.

CTVI generated with CNN denoising showed higher voxel-level correlation compared to other methods.

Median filtering and noise-only CTVI performed worse than CNN-based approaches in preserving ventilation accuracy.

Abstract

Previous studies have reported that quantum noise inherently present in CT images hinders the generation of CT-based ventilation image (CTVI), while quantum noise reduction approaches that do not affect CTVI have not yet been reported.

The purpose of this study was to evaluate the impact of noise reduction preprocessing on the accuracy and robustness of CTVI in relation to quantum noise present in CT images.

To reproduce the quantum noise, Gaussian noise (SD: 30, 80, 150 HU) was added to each inhalation and exhalation CT image. CTVIref and CTVInoise was generated from CTref and CTnoise. A median filter and the noise reduction by the CNN were also applied to the CT image, which contained the quantum noise, and CTVImed and CTVIcnn was created in the same manner as CTVIref. We evaluated whether the regions classified as high, middle, or low in CTVIref were accurately represented as high, middle, or low in CTVInoise, CTVImed and CTVIcnn. Additionally, to evaluate the ventilation function of each voxel, we compared two-dimensional histograms of CTVIref, CTVInoise, CTVImed and CTVIcnn.

Cohen's kappa coefficient and Spearman's correlation were used to assess the agreement between CTVIref and each of the following: CTVInoise, CTVImed, and CTVIcnn.

CTVIcnn significantly improved categorical consistency and voxel-level correlation of CTVI, particularly under high-noise conditions (150 HU), outperforming both CTVInoise and CTVImed.

CNN-based denoising effectively improved the accuracy and robustness of CTVI under quantum noise.

Figures