fastMRI

fastMRI数据集是由Facebook AI Research和NYU School of Medicine合作创建的大型MRI数据集，包含8344个体积的原始MR测量和临床MR图像。该数据集旨在用于训练和评估机器学习方法在MR图像重建中的应用。数据集包括多种类型的数据，如原始多线圈k空间数据、模拟单线圈k空间数据、地面实况图像和DICOM图像。fastMRI数据集的应用领域包括加速MRI成像，旨在减少医疗成本、减轻患者压力，并使MRI在目前因速度或成本限制而不可行的应用中成为可能。

The fastMRI dataset is a large-scale MRI dataset co-created by Facebook AI Research and NYU School of Medicine, containing 8,344 raw MR measurements and clinical MR images across individual volumes. It is designed for training and evaluating machine learning methods applied to MR image reconstruction. The dataset includes multiple types of data, such as raw multi-coil k-space data, simulated single-coil k-space data, ground-truth images, and DICOM images. The application scenarios of the fastMRI dataset cover accelerated MRI imaging, with the goals of reducing medical costs, alleviating patient burden, and making MRI feasible in applications that were previously infeasible due to speed or cost limitations.

搜集汇总

背景与挑战

背景概述

fastMRI数据集是由Facebook AI Research和NYU School of Medicine合作创建的大型MRI数据集，包含8344个体积的原始MR测量和临床MR图像，旨在训练和评估机器学习方法在MR图像重建中的应用。数据集涵盖多种数据类型，如原始多线圈k空间数据和地面实况图像，主要应用于加速MRI成像，以降低医疗成本并扩展MRI的可行性。