Chen Yao, Decker Savannah M, Bruza Petr, Gladstone David J, Jarvis Lesley A, Pogue Brian W, Samkoe Kimberley S, Zhang Rongxiao
Thayer School of Engineering, Dartmouth College, Hanover, New Hampshire.
Geisel School of Medicine, Dartmouth College, Hanover, New Hampshire.
ArXiv. 2025 Aug 19:arXiv:2409.05680v2.
Accurate patient positioning is crucial for precise radiation therapy dose delivery, as errors in positioning can profoundly influence treatment outcomes. This study introduces a novel application for loco-regional tissue deformation tracking via Cherenkov image analysis during fractionated breast cancer radiation therapy. The primary objective of this research was to develop and test an algorithmic method for Cherenkov-based position accuracy quantification, particularly for loco-regional deformations, which do not have an ideal method for quantification during radiation therapy.
Bio-morphological features in the Cherenkov images, such as vessels, were segmented. A rigid/nonrigid combined registration technique was employed to pinpoint both inter- and intrafractional positioning variations. The methodology was tested on an anthropomorphic chest phantom experiment via shifting a treatment couch with known distances and inducing respiratory motion to simulate interfraction setup uncertainties and intrafraction motions, respectively. It was then applied to a data set of fractionated whole breast radiation therapy human imaging (n = 10 patients).
The methodology provided quantified positioning variations comprising 2 components: a global shift determined through rigid registration and a 2-dimensional variation map illustrating loco-regional tissue deformation quantified via nonrigid registration. Controlled phantom testing yielded an average accuracy of 0.83 mm for couch translations up to 20 mm in various directions. Analysis of clinical Cherenkov imaging data from 10 breast cancer patients compared with their first imaged fraction revealed an interfraction setup variation of 3.7 ± 2.4 mm in the global shift and loco-regional deformation up to 3.3 ± 1.9 mm (95th percentile of all regional deformation).
This study introduces the use of Cherenkov visualized bio-morphological features to quantify the global and local variations in patient positioning based on rigid and nonrigid registrations. This new approach demonstrates the feasibility of providing quantitative guidance for inter/intrafraction positioning, particularly for the loco-regional deformations that have been unappreciated in current practice with conventional imaging techniques.
精确的患者定位对于精确的放射治疗剂量传递至关重要，因为定位误差会深刻影响治疗结果。本研究介绍了一种在乳腺癌分次放射治疗期间通过切伦科夫图像分析进行局部组织变形跟踪的新应用。本研究的主要目的是开发并测试一种基于切伦科夫的位置精度量化算法方法，特别是针对局部变形，在放射治疗期间尚无理想的量化方法。
对切伦科夫图像中的生物形态特征（如血管）进行分割。采用刚性/非刚性组合配准技术来确定分次间和分次内的定位变化。该方法通过在拟人化胸部体模实验中以已知距离移动治疗床并诱导呼吸运动，分别模拟分次间设置不确定性和分次内运动来进行测试。然后将其应用于分次全乳放射治疗人体成像数据集（n = 10例患者）。
该方法提供了量化的定位变化，包括两个部分：通过刚性配准确定的全局位移和通过非刚性配准量化局部组织变形的二维变化图。在体模控制测试中，对于在各个方向上最大20毫米的治疗床平移，平均精度为0.83毫米。对10例乳腺癌患者的临床切伦科夫成像数据与其首次成像分次进行比较，结果显示全局位移的分次间设置变化为3.7±2.4毫米，局部变形可达3.3±1.9毫米（所有区域变形的第95百分位数）。
本研究介绍了利用切伦科夫可视化生物形态特征，基于刚性和非刚性配准来量化患者定位的全局和局部变化。这种新方法证明了为分次间/分次内定位提供定量指导的可行性，特别是对于当前传统成像技术实践中未被重视的局部变形。
