LoDoPaB-CT, a benchmark dataset for low-dose computed tomography reconstruction

Johannes Leuschner,Maximilian Schmidt,Peter Maass,Daniel Otero Baguer

doi:10.1038/s41597-021-00893-z

Abstract

Deep learning approaches for tomographic image reconstruction have become very effective and have been demonstrated to be competitive in the field. Comparing these approaches is a challenging task as they rely to a great extent on the data and setup used for training. With the Low-Dose Parallel Beam (LoDoPaB)-CT dataset, we provide a comprehensive, open-access database of computed tomography images and simulated low photon count measurements. It is suitable for training and comparing deep learning methods as well as classical reconstruction approaches. The dataset contains over 40000 scan slices from around 800 patients selected from the LIDC/IDRI database. The data selection and simulation setup are described in detail, and the generating script is publicly accessible. In addition, we provide a Python library for simplified access to the dataset and an online reconstruction challenge. Furthermore, the dataset can also be used for transfer learning as well as sparse and limited-angle reconstruction scenarios.

Highlights

Background & SummaryTomographic image reconstruction is an extensively studied field
To validate the usability of the proposed dataset for machine learning approaches, we provide reference reconstructions and quantitative results for the standard www.nature.com/scientificdata filtered back-projection (FBP) and a learned post-processing method (FBP + U-Net)
The peak signal-to-noise ratio (PSNR) and the structural similarity[45] (SSIM) are two standard image quality metrics often used in computed tomography (CT) applications[42,47]

Summary

Introduction

Background & SummaryTomographic image reconstruction is an extensively studied field. One popular imaging modality in clinical and industrial applications is computed tomography (CT). It allows for the non-invasive acquisition of the inside of an object or the human body. Analytical methods, like filtered back-projection (FBP) or iterative reconstruction (IR) techniques, are used for this task. These methods are the gold standard in the presence of enough high-dose/low-noise measurements. Analytical methods are only able to use very limited prior information. Machine learning approaches are able to learn underlying distributions and typical image features, which constitute a much larger and flexible prior. Recent image reconstruction approaches involving machine learning, in particular deep learning (DL), have been developed and demonstrated to be very competitive[2,3,4,5,6,7,8]

Objectives

Methods

Results