SwinCCIR: An end-to-end deep network for Compton camera imaging reconstruction

2025-12-27

AI Summary

Compton camera reconstruction suffers from intrinsic artifacts, geometric distortions, and system calibration errors, severely limiting imaging quality. This paper proposes the first end-to-end deep learning framework that directly reconstructs the spatial distribution of radioactive sources from list-mode events, bypassing the conventional two-stage paradigm. Its core innovation lies in the first application of the Swin Transformer to Compton imaging, integrated with event encoding and a transposed convolutional generator to enable physics-informed mapping from the event domain to the image domain—implicitly compensating for model inaccuracies and system imperfections. Trained jointly on simulated and experimental data, the method substantially suppresses artifacts and improves both spatial resolution and quantitative accuracy, outperforming state-of-the-art iterative and supervised learning approaches. It demonstrates strong potential for clinical and field-deployable applications.

Technology Category

• Computer Vision: Computational Photography, Image & Video Synthesis
• Machine Learning: Calibration & Uncertainty Quantification
• Search and Optimization: Learning to Search

Application Category

• Security and Privacy: Data transparency and provenance
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation

Abstract

Compton cameras (CCs) are a kind of gamma cameras which are designed to determine the directions of incident gammas based on the Compton scatter. However, the reconstruction of CCs face problems of severe artifacts and deformation due to the fundamental reconstruction principle of back-projection of Compton cones. Besides, a part of systematic errors originated from the performance of devices are hard to remove through calibration, leading to deterioration of imaging quality. Iterative algorithms and deep-learning based methods have been widely used to improve reconstruction. But most of them are optimization based on the results of back-projection. Therefore, we proposed an end-to-end deep learning framework, SwinCCIR, for CC imaging. Through adopting swin-transformer blocks and a transposed convolution-based image generation module, we established the relationship between the list-mode events and the radioactive source distribution. SwinCCIR was trained and validated on both simulated and practical dataset. The experimental results indicate that SwinCCIR effectively overcomes problems of conventional CC imaging, which are expected to be implemented in practical applications.

Problem

• Reconstructs Compton camera images with reduced artifacts
• Addresses systematic errors from device performance limitations
• Establishes direct event-to-distribution mapping via deep learning

Innovation

• End-to-end deep learning framework for Compton camera imaging
• Swin-transformer blocks for event-to-image relationship
• Transposed convolution module generates radioactive source distribution