[Paper Review] Detecting micro fractures with X-ray computed tomography.

Dongwon Lee, Nikolaos Karadimitriou+2 more|arXiv (Cornell University)|Mar 23, 2021

Mineral Processing and Grinding78 references4 citations

TL;DR

This study evaluates five segmentation methods—three conventional and two machine learning-based—for detecting microfractures in X-ray computed tomography (XRCT) data of Carrara marble. The 2D U-Net deep learning model outperformed others in segmentation accuracy and processing speed, demonstrating superior performance despite memory constraints.

ABSTRACT

Studying porous rock materials with X-Ray Computed Tomography (XRCT) has been established as a standard procedure for the non-destructive visualization of flow and transport in opaque porous media. Despite the recent advances in the field of XRCT, some challenges still remain due to the inherent noise and imaging artefacts in the produced data. These issues become even more profound when the objective is the identification of fractures, and/or fracture networks. The challenge is the limited contrast between the regions of interest and the neighboring areas. This limited contrast can mostly be attributed to the minute aperture of the fractures. In order to overcome this challenge, it has been a common approach to apply digital image processing, such as filtering, to enhance the signal-to-noise ratio. Additionally, segmentation methods based on threshold-/morphology schemes can be employed to obtain enhanced information from the features of interest. However, this workflow needs a skillful operator to fine-tune its input parameters, and the required computation time significantly increases due to the complexity of the available methods, and the large volume of the data-set. In this study, based on a data-set produced by the successful visualization of a fracture network in Carrara marble with XRCT, we present the segmentation results from a number of segmentation methods. Three conventional and two machine-learning-based methods are evaluated. The segmentation results from all five methods are compared to each other in terms of segmentation quality and time efficiency. Due to memory limitations, and in order to accomplish a fair comparison, all the methods are employed in a 2D scheme. The output of the 2D U-net model, which is one of the adopted machine-learning-based segmentation methods, shows the best performance regarding the quality of segmentation and the required processing time.

Motivation & Objective

• To address the challenge of low contrast between microfractures and surrounding matrix in XRCT data of porous rocks.
• To evaluate the performance of conventional and machine learning-based segmentation techniques in detecting microfractures.
• To compare segmentation quality and computational efficiency across multiple methods under memory constraints.
• To determine whether deep learning approaches can outperform traditional thresholding and morphological methods in microfracture detection.

Proposed method

• A 2D implementation of the U-Net convolutional neural network was applied for semantic segmentation of microfractures in XRCT data.
• Three conventional segmentation methods based on thresholding and morphological operations were used for comparison.
• All methods were applied to a 2D slice of XRCT data from a fracture network in Carrara marble to ensure fair computational comparison.
• Segmentation quality was assessed by visual and quantitative comparison against ground truth, though no explicit metrics are provided.
• Processing time was measured to evaluate computational efficiency across methods.
• The study focused on 2D processing due to memory limitations, even though 3D data was available.

Experimental results

Research questions

• RQ1 Can deep learning-based segmentation outperform traditional thresholding and morphological methods in detecting microfractures in noisy XRCT data?
• RQ2 How does the 2D U-Net model compare to conventional methods in terms of segmentation accuracy for microfractures?
• RQ3 What is the trade-off between segmentation quality and processing time across different methods?
• RQ4 To what extent do image noise and low contrast between fractures and matrix affect segmentation performance?

Key findings

• The 2D U-Net model achieved the highest segmentation quality among all evaluated methods.
• The 2D U-Net model required less processing time than the other methods, demonstrating superior time efficiency.
• Conventional methods based on thresholding and morphology required extensive parameter tuning and were less effective due to low contrast and noise.
• Despite using a 2D implementation, the U-Net model produced segmentation results that were qualitatively superior to those of conventional approaches.
• Memory limitations necessitated a 2D approach, but the U-Net still outperformed others in both accuracy and speed under these constraints.
• The study confirms that deep learning methods are more robust than traditional image processing pipelines for microfracture detection in noisy XRCT data.