Synapse
⌘+K
Synapse
PulseExploreClubsResearchersJournals
Instagram
HomeClubsExplore
November 1, 2025SPE Journal

A Segmentation-Independent Workflow for Digital Rock Property Prediction Using Pore-Structure-Preserving Generative Adversarial Networks and Synthetic Data: Porosity Case Study

View Full Paper
Ask AI
Bookmark
Share

Authors

ATAhmed TemaniYFYIN Feng

Discussion

Loading...

Member takes

Overview

This approach demonstrates effective property prediction in rock samples using machine learning and synthetic data, highlighting reduced segmentation bias.

Key Points

  • Excellent predictive performance achieved with mean absolute error of 0.56% on unseen rock CT slices, ensuring robust accuracy.
  • Machine learning models employ synthetic data to predict rock properties directly from CT images, bypassing segmentation issues.
  • Analysis used convolutional neural networks for property prediction, validating the method on diverse sandstone formations.
  • This workflow addresses sampling bias and generalization limitations, suggesting scalability for various rock properties.

Cite This Study

Temani et al. (2025) studied this question.

synapsesocial.com/papers/69254371c0ce034ddc35899ehttps://doi.org/10.2118/231405-pa
View Full Paper
Ask AI
Bookmark
Share

Also Consider

Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context:

  1. 1Machine Learning Assisted Prediction of Porosity and Related Properties Using Digital Rock Images2024 · 6 citations
  2. 2Rapid Rock Properties Estimation from Micro-CT Images by Deep Leaning2024
  3. 3PCP-GAN: Property-Constrained Pore-scale image reconstruction via conditional Generative Adversarial Networks2026
  4. 4Using machine learning to discriminate between mineral phases and pore morphologies in carbonate systems2024
  5. 5Conditional Diffusion Models With Integrated Porosity Fusion for 3D Digital Rock Reconstruction2025