Deep learning based Compton backscatter imaging with scattered X-ray spectrum data: A Monte Carlo study

Abstract

Compton backscatter imaging (CBI) is a non-destructive testing (NDT) technique with the Compton effect. In CBI, the radiation source and detector are positioned on the same side of the object, which means that the imaging is almost not limited by the size of the object. The characteristic of single-sided imaging gives CBI substantial potential for applications in specific scenarios. In field archaeology, CBI can rapidly acquire subsurface images with the millimeter-level resolution, effectively enhancing archaeological excavation efficiency while better protecting artifacts. Therefore, CBI offers substantial potential in archaeological excavation. To address the problem of severe noise in CBI, we employed a scintillation detector to receive backscattered X-rays. Transformers directly reconstruct multi-channel projections to swiftly and accurately obtain precise reconstruction images. Monte Carlo simulation experiments confirm this method’s suitability for detecting a variety of common archaeological materials. Experimental results indicate that compared with count mode, multi-channel projections can effectively improve the reconstruction quality of CBI. Moreover, CBI can clearly image various materials of buried artifacts, proving the method’s versatility and practical value in NDT field.