Capability enhancement of the X-ray micro-tomography system via ML-assisted approaches

TL;DR

The paper tackles ring artifacts in X-ray micro-tomography (SXMCT) images by introducing a UNet-based convolutional neural network trained on synthetically generated ring-artifact data. It contrasts this DL approach with traditional non-ML filters using $SSIM$ and $MSE$, demonstrating superior artifact removal and image fidelity. A comprehensive ablation study shows that a six-encoder UNet with transposed-convolution upsampling achieves the best reconstruction quality ($SSIM$ ≈ 0.9523, $MSE$ ≈ 0.0011), while a synthetic data strategy comprising 25 concentric ring masks over 101 real images yields a diverse training set of 2525 pairs. The work highlights end-to-end learning as an effective, efficient alternative for RAF removal in MicroCT, with potential to improve both qualitative interpretation and quantitative analysis without hardware changes.

Abstract

Ring artifacts in X-ray micro-CT images are one of the primary causes of concern in their accurate visual interpretation and quantitative analysis. The geometry of X-ray micro-CT scanners is similar to the medical CT machines, except the sample is rotated with a stationary source and detector. The ring artifacts are caused by a defect or non-linear responses in detector pixels during the MicroCT data acquisition. Artifacts in MicroCT images can often be so severe that the images are no longer useful for further analysis. Therefore, it is essential to comprehend the causes of artifacts and potential solutions to maximize image quality. This article presents a convolution neural network (CNN)-based Deep Learning (DL) model inspired by UNet with a series of encoder and decoder units with skip connections for removal of ring artifacts. The proposed architecture has been evaluated using the Structural Similarity Index Measure (SSIM) and Mean Squared Error (MSE). Additionally, the results are compared with conventional filter-based non-ML techniques and are found to be better than the latter.

Figures (9)

• Figure 1: Experimantal micro-ct image containing ring artifacts
• Figure 2: (a) Flowchart for ring artifact simulation and its removal, where D1 is the set of micro-ct images without ring artifacts, and D2 is the set of synthetic images having ring artifacts. (b) Flowchart of Non-ML method.
• Figure 3: Synthetic Data Generation Process
• Figure 4: Important features for ring artifacts creation
• Figure 5: Samples of synthetic images with ring artifacts