Real-time markerless tumour tracking with patient-specific deep learning using a personalised data generation strategy: proof of concept by phantom study.
- 作者列表："Takahashi W","Oshikawa S","Mori S
OBJECTIVE:For real-time markerless tumour tracking in stereotactic lung radiotherapy, we propose a different approach which uses patient-specific deep learning (DL) using a personalised data generation strategy, avoiding the need for collection of a large patient data set. We validated our strategy with digital phantom simulation and epoxy phantom studies. METHODS:We developed lung tumour tracking for radiotherapy using a convolutional neural network trained for each phantom's lesion by using multiple digitally reconstructed radiographs (DRRs) generated from each phantom's treatment planning four-dimensional CT. We trained tumour-bone differentiation using large numbers of training DRRs generated with various projection geometries to simulate tumour motion. We solved the problem of using DRRs for training and X-ray images for tracking using the training DRRs with random contrast transformation and random noise addition. RESULTS:We defined adequate tracking accuracy as the percentage frames satisfying <1 mm tracking error of the isocentre. In the simulation study, we achieved 100% tracking accuracy in 3 cm spherical and 1.5×2.25×3 cm ovoid masses. In the phantom study, we achieved 100 and 94.7% tracking accuracy in 3 cm and 2 cm spherical masses, respectively. This required 32.5 ms/frame (30.8 fps) real-time processing. CONCLUSIONS:We proved the potential feasibility of a real-time markerless tumour tracking framework for stereotactic lung radiotherapy based on patient-specific DL with personalised data generation with digital phantom and epoxy phantom studies. ADVANCES IN KNOWLEDGE:Using DL with personalised data generation is an efficient strategy for real-time lung tumour tracking.
目的: 对于立体定向肺放疗中的实时无标记肿瘤跟踪，我们提出了一种不同的方法，该方法使用患者特异性深度学习 (DL)，使用个性化数据生成策略，避免需要收集大型患者数据集。我们通过数字体模模拟和环氧体体模研究验证了我们的策略。 方法: 我们开发了用于放疗的肺部肿瘤跟踪，使用卷积神经网络，通过使用多个数字重建的射线照片 (DRRs) 对每个体模病变进行训练。从每个体模的治疗计划生成的四维CT。我们使用各种投影几何形状生成的大量训练drr来训练肿瘤-骨分化，以模拟肿瘤运动。我们解决了使用随机对比度变换和随机噪声添加的训练DRRs进行训练和x射线图像跟踪的问题。 结果: 我们将足够的跟踪精度定义为满足等中心 <1 mm跟踪误差的百分比帧。在仿真研究中，我们在 3 cm球形和 100% × 1.5 × 3 cm卵圆形质量中实现了 2.25 的跟踪精度。在体模研究中，我们分别在 100 和 2 cm球形质量中实现了 94.7% 和 3厘米的跟踪精度。这需要 32.5 ms/帧 (30.8 fps) 实时处理。 结论: 我们证明了基于患者特异性DL的立体定向肺放疗的实时无标记肿瘤跟踪框架的潜在可行性，并通过数字体模和环氧体模研究进行个性化数据生成。 知识进展: 使用具有个性化数据生成的DL是实时肺部肿瘤跟踪的有效策略。
METHODS::Pulmonary artery sling is a rare congenital anomaly of the origin and course of the left pulmonary artery. Patients with this condition typically present with respiratory failure in young infancy, and asymptomatic cases are uncommon. We describe the case of an adult patient with a lung adenocarcinoma of the right upper lobe, extending into the hilum and superior mediastinum, and with a previously unknown pulmonary artery sling anomaly. The local invasiveness of the tumor and the peculiar vascular anatomy contributed to a unique surgical scenario, wherein multiple reconstructive procedures were required.
METHODS::Patients with idiopathic pulmonary fibrosis (IPF) have higher risk of developing lung cancer, for example, squamous cell carcinoma (SCC), and show poor prognosis, while the molecular basis has not been fully investigated. Here we conducted DNA methylome analysis of lung SCC using 20 SCC samples with/without IPF, and noncancerous lung tissue samples from smokers/nonsmokers, using Infinium HumanMethylation 450K array. SCC was clustered into low- and high-methylation epigenotypes by hierarchical clustering analysis. Genes hypermethylated in SCC significantly included genes targeted by polycomb repressive complex in embryonic stem cells, and genes associated with Gene Ontology terms, for example, "transcription" and "cell adhesion," while genes hypermethylated specifically in high-methylation subgroup significantly included genes associated with "negative regulation of growth." Low-methylation subgroup significantly correlated with IPF (78%, vs. 17% in high-methylation subgroup, p = 0.04), and the correlation was validated by additional Infinium analysis of SCC samples (n = 44 in total), and data from The Cancer Genome Atlas (n = 390). The correlation between low-methylation subgroup and IPF was further validated by quantitative methylation analysis of marker genes commonly hypermethylated in SCC (HOXA2, HOXA9 and PCDHGB6), and markers specifically hypermethylated in high-methylation subgroup (DLEC1, CFTR, MT1M, CRIP3 and ALDH7A1) in 77 SCC cases using pyrosequencing (p = 0.003). Furthermore, low-methylation epigenotype significantly correlated with poorer prognosis among all SCC patients, or among patients without IPF. Multivariate analysis showed that low-methylation epigenotype is an independent predictor of poor prognosis. These may suggest that lung SCC could be stratified into molecular subtypes with distinct prognosis, and low-methylation lung SCC that significantly correlates with IPF shows unfavorable outcome.
METHODS::The role of Fyn-related kinase (FRK) in malignant tumors remains controversial. Our study investigated the function of FRK in lung cancer. Immunohistochemistry staining and generating a knockout of FRK by CRISPR/Cas9 in H1299 (FRK-KO-H1299) cells were strategies used to explore the role of FRK. Immunohistochemistry staining indicated that FRK expression was elevated in 223 lung cancer tissues compared to 26 distant normal lung tissues. FRK contributed to poor survival status in lung cancer patients and acted as a predictor for poor prognosis of lung cancer. Knockout of FRK by CRISPR/Cas9 markedly inhibited proliferation, invasion, colony formation and epithelial-mesenchymal transition (EMT) process in the lung cancer cell line H1299. Further exploration indicated that FRK-KO damaged the stemness phenotype of H1299 by inhibiting CD44 and CD133 expression. Seahorse detection and a U-13 C flux assay revealed that FRK-KO induced metabolism reprogramming by inhibiting the Warburg effect and changing the energy type in H1299 cells. Epidermal growth factor stimulation recovered the expression of FRK and biological functions, metabolic reprogramming and stemness phenotype of H1299 cells. FRK plays an oncogenic role in lung cancer cells via a novel regulation mechanism of enhancing the stemness of H1299 cells by inducing metabolism reprogramming, which finally promotes EMT and metastasis. Our study also indicates that FRK could be used as a potential therapeutic target for drug development.