Artificial intelligence-driven concrete computed tomography data analysis in engineering: Integrating ChatGPT with professional visualization software

Computed tomography (CT) is widely used for non-destructive testing of engineering materials. But traditional CT image processing suffers from high technical barriers, low efficiency, and over-reliance on subjective expertise. To address these limitations, this study proposes an integrated framework combining artificial intelligence (AI) and professional visualization software (PVS) to enhance processing efficiency across multiple domains, which focuses on CT image processing of concrete structures obtained from field drilling, exploring the combined application of ChatGPT-based generative techniques and PVS in tasks such as image denoising and enhancement, segmentation, and quality assessment. Using quantitative metrics including peak signal-to-noise ratio (PSNR) and grayscale distribution, we systematically evaluated the performance of ChatGPT in CT data processing and compared the outcomes of using PVS alone, ChatGPT alone, and their integration. The results demonstrate that: (1) ChatGPT alone achieves superior denoising and enhancement performance, with ring artifact removal PSNR = 33.26 dB, denoising PSNR = 56.27 dB. (2) ChatGPT-generated scripts exhibit superior performance in ring artifact removal and denoising. (3) Compared to PVS alone, ChatGPT-PVS reduces ring artifact removal time from 46.27 min to 5.43 min (88.26 % reduction) with PSNR increased from 28.29 dB to 33.26 dB; denoising time from 75.09 min to 6.38 min (91.50 % reduction) with PSNR improved from 28.58 dB to 51.97 dB; quality assessment time from 24.05 min to 17.49 min (27.28 % reduction) with porosity consistency closer to the verification value. (4) Despite its strengths in ring artifact and noise reduction, ChatGPT-based scripts show limitations in the intelligent identification and segmentation of complex multiphase structures. This study provides a comprehensive investigation into the application of dialogue-based generative AI in industrial CT data processing, offering a practical reference for structural inspection and health monitoring. It simplifies CT image processing workflows, lowers the technical thresholds, and lays a foundation for enhanced efficiency in the engineering sector.