arXiv subjects

Staffan Holmin

Publications and source records attributed to Staffan Holmin.

3 recordsLinked to original sources

arXiv2026

Unpaired Deep Learning Synthesis of Photon-Counting CT Material Basis Maps from Non-contrast Energy-Integrating Abdominal CT Images

Photon-counting Computed Tomography (PCCT) is the most advanced Computed Tomography (CT) technology, offering significant improvements in image quality and diagnostic capabilities. However, since PCCT has only recently been adopted in the clinc, there are no publicly available PCCT image datasets for study. We therefore aim to synthesize PCCT spectral material-basis images from publicly available EID CT images. We propose a two-step deep learning model designed to synthesize photon-counting spectral material basis images from public Energy-Integrating Detector (EID) CT images. In the first step, we use a Denoising Diffusion Implicit Model (DDIM) to generate EID CT images from PCCT images. In the second step we use a U-Net with a Domain-Adversarial Neural Network to predict water and iodine maps from generated EID CT images. We also reconstruct basis images and virtual monoenergetic images (VMIs) from the predicted material-basis maps for evaluation. We evaluated the generated water and iodine maps as well as the 40 and 70 keV PCCT images in terms of Hounsfield Unit accuracy, modulation transfer function and noise power spectrum as well as qualitative image appearance. The reconstructed 40 and 70 keV PCCT images exhibit higher spatial resolution while preserving the anatomical structures and textures of the original EID CT images, thereby demonstrating the feasibility of the proposed approach. The proposed framework provides a feasible approach for synthesizing PCCT spectral material-basis images from conventional EID CT without requiring paired images. This method has the potential to provide large sets of synthetic training and evaluation data for PCCT algorithm development in data-limited environments.

physics.med-ph↗

arXiv2024

Noise suppression in photon-counting CT using unsupervised Poisson flow generative models

Deep learning has proven to be important for CT image denoising. However, such models are usually trained under supervision, requiring paired data that may be difficult to obtain in practice. Diffusion models offer unsupervised means of solving a wide range of inverse problems via posterior sampling. In particular, using the estimated unconditional score function of the prior distribution, obtained via unsupervised learning, one can sample from the desired posterior via hijacking and regularization. However, due to the iterative solvers used, the number of function evaluations (NFE) required may be orders of magnitudes larger than for single-step samplers. In this paper, we present a novel image denoising technique for photon-counting CT by extending the unsupervised approach to inverse problem solving to the case of Poisson flow generative models (PFGM)++. By hijacking and regularizing the sampling process we obtain a single-step sampler, that is NFE=1. Our proposed method incorporates posterior sampling using diffusion models as a special case. We demonstrate that the added robustness afforded by the PFGM++ framework yields significant performance gains. Our results indicate competitive performance compared to popular supervised, including state-of-the-art diffusion-style models with NFE=1 (consistency models), unsupervised, and non-deep learning-based image denoising techniques, on clinical low-dose CT data and clinical images from a prototype photon-counting CT system developed by GE HealthCare.

physics.med-ph↗

arXiv2023

PPFM: Image denoising in photon-counting CT using single-step posterior sampling Poisson flow generative models

Diffusion and Poisson flow models have shown impressive performance in a wide range of generative tasks, including low-dose CT image denoising. However, one limitation in general, and for clinical applications in particular, is slow sampling. Due to their iterative nature, the number of function evaluations (NFE) required is usually on the order of $10-10^3$, both for conditional and unconditional generation. In this paper, we present posterior sampling Poisson flow generative models (PPFM), a novel image denoising technique for low-dose and photon-counting CT that produces excellent image quality whilst keeping NFE=1. Updating the training and sampling processes of Poisson flow generative models (PFGM)++, we learn a conditional generator which defines a trajectory between the prior noise distribution and the posterior distribution of interest. We additionally hijack and regularize the sampling process to achieve NFE=1. Our results shed light on the benefits of the PFGM++ framework compared to diffusion models. In addition, PPFM is shown to perform favorably compared to current state-of-the-art diffusion-style models with NFE=1, consistency models, as well as popular deep learning and non-deep learning-based image denoising techniques, on clinical low-dose CT images and clinical images from a prototype photon-counting CT system.

eess.IV↗