Low-Dose CT With a Residual Encoder-Decoder Convolutional Neural Network

TLDR

Low‑dose CT is pursued to reduce X‑ray radiation risk, yet existing vendor‑specific sinogram filtering and iterative reconstruction require raw data, and image‑domain approaches struggle to suppress noise while preserving detail. The study aims to create a residual encoder‑decoder convolutional neural network (RED‑CNN) for low‑dose CT imaging. RED‑CNN integrates an autoencoder, deconvolution layers, and shortcut connections to process reconstructed images directly. Patch‑based training of RED‑CNN yields performance competitive with state‑of‑the‑art methods, achieving superior noise suppression, structural preservation, and lesion detection in both simulated and clinical cases.

Abstract

Given the potential risk of X-ray radiation to the patient, low-dose CT has attracted a considerable interest in the medical imaging field. Currently, the main stream low-dose CT methods include vendor-specific sinogram domain filtration and iterative reconstruction algorithms, but they need to access raw data, whose formats are not transparent to most users. Due to the difficulty of modeling the statistical characteristics in the image domain, the existing methods for directly processing reconstructed images cannot eliminate image noise very well while keeping structural details. Inspired by the idea of deep learning, here we combine the autoencoder, deconvolution network, and shortcut connections into the residual encoder-decoder convolutional neural network (RED-CNN) for low-dose CT imaging. After patch-based training, the proposed RED-CNN achieves a competitive performance relative to the-state-of-art methods in both simulated and clinical cases. Especially, our method has been favorably evaluated in terms of noise suppression, structural preservation, and lesion detection.