Thorax x‐ray and CT interventional dataset for nonrigid 2D/3D image registration evaluation

Abstract

Purpose The aim of this paper is to provide a novel, publicly available standard image dataset with a useful evaluation framework for assessing nonrigid two‐/three‐dimensional (2D/3D) registration algorithms. Acquisition and validation methods A pig lung model was used to obtain the image dataset. Inflated with different amounts of oxygen, a sequence of 3D volume data was acquired with computed tomography (CT), which ideally simulated different respiratory phases. With the model inflated and kept in certain states, 3D CT, 2D CT scout image and 2D x‐ray were acquired for the same respiratory phases, making them suitable to establish the evaluation dataset for 2D/3D registration algorithms. A total of 120 well‐distributed landmarks in every 3D volume were manually annotated and checked by several radiologists using semi‐automatic software to generate the dataset. Data format and usage notes All 3D image data were stored in both DICOM and ITK Meta format, and 2D image data were stored in DICOM format. A total of 120 landmarks were manually annotated for each 3D image. Among these landmarks, eight landmarks located on large branch of the bronchial tree were also annotated in 2D images. The landmark coordinates were stored in a text file. The detailed usage including a standard evaluation framework for the proposed dataset is also provided. The dataset can be downloaded from the Zenodo repository (https://doi.org/10.5281/zenodo.997887). Potential applications Our standard dataset was acquired with advanced clinical imaging devices and is quite suitable for quantitatively evaluating state‐of‐art, nonrigid 2D/3D registration algorithms.

A 15-Gbps BiCMOS XNOR gate for fast recognition of COVID-19 in binarized neural networks

“…On the contrary, CT scans generates 3-dimensional, high-quality, detailed images of the body. It is considered a more powerful and sophisticated tool that scans a 360-degree image of internal organs [18], [19]. However, the output image of a CT is much larger than an X-ray.…”

A new 2D‐3D registration gold‐standard dataset for the hip joint based on uncertainty modeling

“…To date, only few gold standard datasets are publicly available for the validation of 2D‐3D registration for application in orthopedics. They include sets of CT, MR volumes, and x‐ray images of human cadaveric spines 26 , 27 and of a fresh porcine cadaver head 28 and lungs 29 as well as a simulated dataset of CT and digitally reconstructed radiographs (DRRs) of human pelvis and vertebrae from the Visible Human Project. 30 Using synthetic DRR images provides an exact known ground truth but usually results in non‐fully realistic x‐ray images (e.g., absence of x‐ray scattering or image noise).…”

Towards Automated Spine Mobility Quantification: A Locally Rigid CT to X-ray Registration Framework

“…This reduces the complexity of this ill-posed problem but limits the potential accuracy and applications, as it cannot capture local deformations of soft tissue related to different positions of the subject. Non-rigid advances have been mostly focused on other body parts, for example, on lung movement [12], and might be unrealistic in rigid anatomical structures such as bones.…”