[论文解读] Head and Neck Tumor Segmentation from [18F]F-FDG PET/CT Images Based on 3D Diffusion Model

Yafei Dong, Kuang Gong|arXiv (Cornell University)|Jan 31, 2024

Radiomics and Machine Learning in Medical Imaging被引用 1

一句话总结

本文提出一种3D扩散模型，结合[18F]F-FDG PET与CT影像，以提升头颈部肿瘤分割性能。该模型采用3D U-Net架构对拼接的PET、CT及高斯噪声体积进行去噪，实现0.739的平均Dice分数，优于2D扩散模型（0.669）与单模态方法（平均Dice < 0.570），证明其在多模态3D医学图像分割中具备更优的准确性与鲁棒性。

ABSTRACT

Head and neck (H&N) cancers are among the most prevalent types of cancer worldwide, and [18F]F-FDG PET/CT is widely used for H&N cancer management. Recently, the diffusion model has demonstrated remarkable performance in various image-generation tasks. In this work, we proposed a 3D diffusion model to accurately perform H&N tumor segmentation from 3D PET and CT volumes. The 3D diffusion model was developed considering the 3D nature of PET and CT images acquired. During the reverse process, the model utilized a 3D UNet structure and took the concatenation of PET, CT, and Gaussian noise volumes as the network input to generate the tumor mask. Experiments based on the HECKTOR challenge dataset were conducted to evaluate the effectiveness of the proposed diffusion model. Several state-of-the-art techniques based on U-Net and Transformer structures were adopted as the reference methods. Benefits of employing both PET and CT as the network input as well as further extending the diffusion model from 2D to 3D were investigated based on various quantitative metrics and the uncertainty maps generated. Results showed that the proposed 3D diffusion model could generate more accurate segmentation results compared with other methods. Compared to the diffusion model in 2D format, the proposed 3D model yielded superior results. Our experiments also highlighted the advantage of utilizing dual-modality PET and CT data over only single-modality data for H&N tumor segmentation.

研究动机与目标

• 开发一种用于[18F]F-FDG PET/CT影像中头颈部肿瘤精确自动分割的3D扩散模型。
• 评估同时使用PET与CT模态在肿瘤分割中的优势，利用其互补的诊断信息。
• 研究3D操作相较于2D操作在建模体积医学影像数据时的性能提升。
• 评估多模态输入对降低不确定性与提升分割可靠性的影响。
• 为将3D扩散模型应用于头颈部癌症以外的其他医学图像分割任务奠定基础。

提出的方法

• 设计了一种3D扩散模型，采用3D U-Net作为去噪网络，执行反向去噪过程。
• 在推理过程中，网络输入为每个反向步骤中PET、CT与高斯噪声体积的拼接。
• 模型在HECKTOR 2021数据集上进行训练，该数据集包含来自五个国际中心的224例口咽癌病例。
• 反向过程在1000个时间步内迭代去噪，从噪声中重建出肿瘤分割掩码。
• 通过标准指标（包括Dice、Hausdorff距离与敏感性）进行定量评估，并与U-Net及基于Transformer的基线方法进行比较。
• 通过生成每个输入的多个预测结果来量化不确定性，从而分析模型的随机性与置信度。

实验结果

研究问题

• RQ1 3D扩散模型在从PET/CT影像分割头颈部肿瘤方面是否优于2D扩散模型？
• RQ2 与单模态输入相比，融合PET与CT模态是否能提升分割准确性？
• RQ3 相较于2D方法，3D体积分模建模在增强特征学习与分割性能方面有何优势？
• RQ4 多模态输入在多大程度上降低了肿瘤分割预测的不确定性？
• RQ5 该3D扩散框架能否推广至其他解剖区域与肿瘤类型？

主要发现

• 所提出的3D扩散模型实现了0.739的平均Dice分数，显著优于其他最先进方法（< 0.726）。
• 3D扩散模型的平均Dice为0.739，而2D版本为0.669，证明了3D体积分模建模的优势。
• 仅使用PET或CT的单模态分割平均Dice分数低于0.570，凸显多模态融合的优势。
• 同时使用PET与CT输入降低了预测不确定性，表现为多次前向传播中结果方差更低。
• 该模型在RTX 8000 GPU上的单例推理时间约为14.7分钟，表明尽管性能优异，但计算成本较高。
• 本研究证实，3D扩散模型在该任务中可生成比传统U-Net与基于Transformer的模型更精确、更可靠的肿瘤分割掩码。