The Head and Neck Lymph Nodes Dataset, Definitive, Non-contrast (RPA-Head-and-Neck-Lymph-Nodes)

The Radiation Planning Assistant (RPA), a suite of fully automated contouring and radiotherapy treatment planning tools, is being developed in the Department of Radiation Physics at MD Anderson Cancer Center in Houston, TX and has 510(k) clearance by the FDA (although not being marketed). The mandatory inputs into the system are a simulation CT scan (approved by radiographer) and a Treatment Plan Order (approved by a radiation oncologist). The RPA then generates a complete radiotherapy treatment plan. The overall goal of the RPA is to improve access to high quality radiotherapy across the world. In this submission, we are providing data used for training a model for deployment in the RPA and/or MD Anderson’s radiation oncology clinic for the Head & Neck Lymph Nodes site. These include data being used to train models for future versions of the RPA. This dataset provides a benchmark for the auto-contouring of lymph node levels in the head and neck. Simulation CTs (non-contrast) and clinical segmentations of 47 (46 unique) patients who retrospectively received head and neck radiotherapy at MD Anderson Cancer Center were curated. All patients received definitive therapy (i.e. no nodal level dissection). We hope that this RPA-based dataset will be beneficial for future studies involving auto segmentation in radiation therapy treatment planning.

放射治疗规划助手（Radiation Planning Assistant，RPA）是一套全自动化轮廓勾画与放射治疗规划工具，由位于美国德克萨斯州休斯顿的MD安德森癌症中心放射物理部开发，已获得美国食品药品监督管理局（Food and Drug Administration，FDA）的510(k)认证（尽管尚未商业化上市）。该系统的强制输入项为经放射技师审核的模拟CT扫描影像，以及经放射肿瘤科医师批准的治疗计划医嘱，随后放射治疗规划助手即可生成完整的放射治疗计划。

该工具的总体目标是提升全球范围内高质量放射治疗的可及性。本次提交的数据集，可用于训练可部署于放射治疗规划助手，或MD安德森癌症中心放射肿瘤科的头颈部淋巴结区域放射治疗规划模型，同时也可用于该工具未来版本的模型训练。本数据集为头颈部淋巴结区域的自动轮廓勾画任务提供了基准评测标准。

研究团队汇编整理了47例（其中46例为独立病例）在MD安德森癌症中心接受过头颈部放射治疗的回顾性患者的无增强模拟CT影像，以及对应的临床分割标注结果。所有患者均接受了根治性治疗（即未行淋巴结清扫术）。

我们期望本基于放射治疗规划助手的数据集，能够为未来放射治疗规划中的自动分割相关研究提供助力。

搜集汇总

背景与挑战

背景概述

该数据集由MD Anderson Cancer Center的Radiation Planning Assistant (RPA)项目提供，包含47名接受头颈部放射治疗患者的模拟CT扫描（非对比增强）和临床分割数据，专门用于训练头颈部淋巴结自动轮廓勾画模型。数据集旨在为放射治疗规划中的自动分割提供基准，以提高全球放射治疗质量，适用于研究和临床部署。