[论文解读] Focal FCN: Towards Biomedical Small Object Segmentation with Limited Training Data

Xiao-Yun Zhou, Mali Shen+3 more|arXiv (Cornell University)|Nov 4, 2017

Aortic aneurysm repair treatments参考文献 1被引用 2

一句话总结

本文提出了一种基于等权重焦点U-Net的自动3D支架移植物形状实例化方法，用于机器人血管内手术，通过在2D fluoroscopy图像中分割五种类别的标记点。该方法实现了0.6943的平均交并比（mIoU）和81.01%的标记点中心位置误差小于1.6 mm，实现了在有限训练数据下的高精度实时3D重建。

ABSTRACT

Robot-assisted Fenestrated Endovascular Aortic Repair (FEVAR) is currently navigated by 2D fluoroscopy which is insufficiently informative. Previously, a semi-automatic 3D shape instantiation method was developed to instantiate the 3D shape of a main, deployed, and fenestrated stent graft from a single fluoroscopy projection in real-time, which could help 3D FEVAR navigation and robotic path planning. This proposed semi-automatic method was based on the Robust Perspective-5-Point (RP5P) method, graft gap interpolation and semi-automatic multiple-class marker center determination. In this paper, an automatic 3D shape instantiation could be achieved by automatic multiple-class marker segmentation and hence automatic multiple-class marker center determination. Firstly, the markers were designed into five different shapes. Then, Equally-weighted Focal U-Net was proposed to segment the fluoroscopy projections of customized markers into five classes and hence to determine the marker centers. The proposed Equally-weighted Focal U-Net utilized U-Net as the network architecture, equally-weighted loss function for initial marker segmentation, and then equally-weighted focal loss function for improving the initial marker segmentation. This proposed network outperformed traditional Weighted U-Net on the class-imbalance segmentation in this paper with reducing one hyper-parameter - the weight. An overall mean Intersection over Union (mIoU) of 0.6943 was achieved on 78 testing images, where 81.01% markers were segmented with a center position error <1.6mm. Comparable accuracy of 3D shape instantiation was also achieved and stated. The data, trained models and TensorFlow codes are available on-line.

研究动机与目标

• 解决在3D支架移植物重建中生物医学小目标分割因训练数据有限带来的挑战。
• 通过消除对手动损失加权的需求，克服标记点分割中的类别不平衡问题。
• 通过用深度学习替代人工标记中心检测，实现在机器人血管内手术中自动3D形状实例化。
• 通过新型损失函数提升在2D荧光透视投影图像中对小而独特的标记点的分割精度。
• 开发一种实用且可部署的解决方案，提供开源数据、模型和代码，以支持临床集成。

提出的方法

• 设计了五种不同的标记点形状，以增强在2D荧光透视图像中视觉区分度和分割精度。
• 采用U-Net作为基础网络架构，用于标记点类别的语义分割。
• 应用等权重损失函数进行初始标记点分割，以简化超参数调优。
• 集成等权重焦点损失函数，以改善代表性不足的标记点类别的分割性能。
• 利用分割后的标记点类别自动确定标记点中心位置，用于3D形状实例化。
• 结合标记点中心检测、移植物间隙插值和RP5P-based 3D重建方法，实现实时3D形状估计。

实验结果

研究问题

• RQ1 自动标记点分割方法是否能降低在3D支架移植物重建中对人工标注的依赖？
• RQ2 等权重焦点损失函数与传统加权损失函数相比，在处理小生物医学目标的类别不平衡问题时表现如何？
• RQ3 基于U-Net的模型在荧光透视图像中，使用有限训练数据时，能否实现高分割精度？
• RQ4 标记点形状设计对分割性能和3D重建精度有何影响？
• RQ5 所提出的方法是否能实现临床可接受的精度（误差小于1.6 mm）以实现实时3D形状实例化？

主要发现

• 等权重焦点U-Net在78张测试图像上实现了0.6943的平均交并比（mIoU），表现出优异的分割性能。
• 所有分割标记点中81.01%的中心位置误差低于1.6 mm，表明其在3D重建中具有高几何精度。
• 与传统加权U-Net相比，该方法在处理类别不平衡问题上表现更优，同时减少了一个超参数。
• 自动标记点中心检测实现了与先前半自动方法相当的3D形状实例化精度。
• 该方法在有限训练数据条件下展现出在实时环境中的鲁棒性，适合临床部署。
• 数据集、训练模型和TensorFlow代码已公开，支持可复现性及后续研究。