通过深度学习模型从增强CT图像合成平扫CT图像的质量分析

目的通过基于卷积神经网络深度学习方法从增强CT合成平扫CT图像, 临床主观和客观评估合成平扫CT图像(DL-SNCT)与金标准平扫CT图像的相似性, 探讨其潜在临床价值。方法同时行常规平扫和增强CT扫描的患者34例, 通过深度学习模型将增强CT图像合成DL-SNCT图像, 以平扫CT图像为金标准, 主观评价DL-SNCT的图像质量(评价指标包括解剖结构清晰度、伪影、噪声、图像结构完整性、图像变形, 均采用4分制);利用配对t检验比较DL-SNCT与金标准平扫CT图像不同血供特点的解剖部位(主动脉、肾脏、肝实质、臀大肌)以及不同强化模式的肝脏病变(肝癌、肝血管瘤、肝转移瘤、肝囊肿)的CT值。结果主观评价上, DL-SNCT图像在伪影、噪声、图像结构完整性、图像变形方面评分都达到4分, 与平扫CT图像评分相一致(P>0.05);在解剖结构清晰度方面评分略低于平扫CT图像[(3.59±0.70)分vs. 4分)], 差异有统计学意义(Z=-2.89, P <0.05)。对于不同解剖部位而言, DL-SNCT图像主动脉、肾脏的CT值显著高于平扫CT图像(t=-12.89、-9.58...     

基于12和16 bit CT图像的金属伪影校正对放疗剂量分布的影响

目的 利用金属伪影去除技术去除基于12 bit和16 bit CT图像中金属植入物伪影,分析其对图像CT值分布和放疗剂量分布的影响。方法 将金属棒插入模体中,CT扫描得到12和16 bit原始CT图像,运用归一化伪影去除法（NMAR）分别对所得到的原始CT图像进行去伪影处理,得到NMAR修正后图像。临床中选取人工股骨头患者CT图像,对其进行同样处理。比较分析各图像伪影去除前后CT值分布。在放疗计划系统中,基于各图像设计放射治疗计划,计算剂量分布,比较分析各图像的剂量分布差异。结果 12 bit图像中金属CT值为3 071 HU,远小于金属实际CT值11 080 HU;16 bit图像中金属CT值为11 098 HU,与实际值很接近。原始CT图像在金属周围含有大量伪影,CT值与参考图像CT值偏差很大;NMAR校正后图像伪影显著减少,CT值与参考图像较接近。NMAR修正后16 bit图像的剂量分布与参考图像最接近,中心轴上最大剂量偏差为1.8%;12 bit图像与参考图像在金属后方剂量差异很大,最大剂量偏差为81.6%。射线穿过原始图像伪影区域后导致剂量分布与参考图像有明显差异,引起最大剂量偏差达21.6%。结论 含有金属植入物时,基于16 bit图像进行NMAR伪影校正可以得到准确的CT值分布,从而得到准确的剂量分布。     

一种基于互信息对比学习由磁共振成像生成头部伪CT图像的方法

目的:比较不同神经网络由磁共振成像(MRI)图像生成伪CT图像的本领,探讨伪CT用于临床放疗计划的可行性。方法:选取29例同时具有计划CT和诊断MRI的脑癌患者,23例用于训练,6例用于测试。分别采用循环生成对抗网络(cycleGAN)、对比学习非配对图像转换网络(CUT)以及本研究提出的改进网络denseCUT由MR...     

基础研究

深度学习在光声图像重建及其影像诊断中的应用。生物医学光子学技术可以提供高对比度的生物组织图像,能够实现高分辨率的三维成像,部分光学成像技术已从实验室发展到临床应用阶段。但临床繁复的工作流程限制了同时满足图像的快速形成和可解释性,这对新型光学成像技术在诊疗过程中的应用提出了新挑战。近年来,深度学习与神经网络在光学图像重建、分割和分类这些问题上的应用受到了广泛关注,本文通过回顾深度学习在光学成像诊断技术在临床应用中的进展情况,如在光声图像重建过程中应用深度学习算法对不完全采样的数据进行处理,实现传统算法不可比拟去除伪影和噪声的重构效果;用深度学习算法实现与资深临床医师准确率相近的血管轮廓分割以及光声断层图像所含血管结构所属的血管层的识别,解释了这些方法如何适用于光学成像过程,并讨论了未来深度学习应用于光学诊疗中的方向和挑战。     

基于深度卷积神经网络的CT图像层间插值方法的初步研究

目的：研究一种基于CT图像层间插值的方法,用于放射治疗过程中的患者摆位验证,从而提高放疗精度。方法：采用一种基于3D卷积和膨胀卷积神经网络（3D CNN-DCNN）算法,利用相邻图像层之间的关联信息重建中间层图像。采用U-Net网络架构,通过编码部分的卷积层、膨胀卷积层、池化层和解码部分的上采样层、卷积层、膨胀卷积层,对CT进行端到端的学习。采集20例患者图像数据,采用留一交叉验证的方法训练验证模型,分别对神经网络和线性插值的预测CT与原始薄层CT进行对照比较。结果：3D CNN-DCNN的平均绝对误差（MAE）为34 HU,远小于线性插值的55 HU。除此之外,骨骼的Dice相似系数（DSC）为0.95,大于线性插值方法的0.89。结论：与传统线性插值方法相比,3D CNN-DCNN算法可以更准确的重建薄层CT,明显改善了插值伪影、图像失真和锯齿状现象。     

深度学习在光声图像重建及其影像诊断中的应用

生物医学光子学技术可以提供高对比度的生物组织图像,能够实现高分辨率的三维成像,部分光学成像技术已从实验室发展到临床应用阶段。但临床繁复的工作流程限制了同时满足图像的快速形成和可解释性,这对新型光学成像技术在诊疗过程中的应用提出了新挑战。近年来,深度学习与神经网络在光学图像重建、分割和分类这些问题上的应用受到了广泛关注,本文通过回顾深度学习在光学成像诊断技术在临床应用中的进展情况,如在光声图像重建过程中应用深度学习算法对不完全采样的数据进行处理,实现传统算法不可比拟去除伪影和噪声的重构效果;用深度学习算法实现与资深临床医师准确率相近的血管轮廓分割以及光声断层图像所含血管结构所属的血管层的识别,解释了这些方法如何适用于光学成像过程,并讨论了未来深度学习应用于光学诊疗中的方向和挑战。     

影像组学与深度学习在肿瘤放疗中的研究进展

影像组学作为一种非侵入性的图像分析方法,能够深度发掘隐藏在医学影像背后的临床信息。深度学习技术的发展将影像组学研究提升到了新的高度,大量研究结果证实了其在肿瘤放疗中的应用价值。笔者从影像组学的研究背景出发,就其在肿瘤放疗中的研究进展进行综述。     

能谱成像技术去除金属伪影的临床价值

目的：研究宝石CT能谱扫描在减少金属伪影方面的临床价值。方法：对31例体内含有金属植入物的受检者行能谱扫描（Gemstone spectral imaging,GSI）,扫描后获得混合能量图像（140kVp）,用能谱分析软件（GSI Viewer）进行分析,以10keV为间距在40~140keV间进行11种不同能量的单能量图像重建,选取最优单能量图像,再行金属伪影消除重建（Metal-Artifacts Reduction System,MARs）,对混合能量图像及能谱图像（110keV单能量图像或单能量＋MARs图像）进行感兴趣区（ROI）SD值的测定,计算出伪影的SD值。并且所有图像均由三位有经验的放射医师采用盲法进行独立评分,按金属伪影对图像质量的影响程度予以记3、2、1、0分（3分为基本无伪影;2分为图像质量较好,有部分伪影;1分为图像伪影较重,尚能观察;0分为伪影很重,图像无法观察）。对所获数据采用SPSS 17.0进行配对t检验分析。结果：在110keV单能量区图像信噪比较高,因此所有图像均于110keV行MARs重建。能谱图像（110keV单能量图像或单能量＋MARs图像）的评分与混合能量图像的评分之间,以及能谱图像组与混合能量图像组金属伪影的SD值之间均存在显著性差异（P=0.000〈0.05）,即能量图像的金属伪影明显降低,图像质量优于混合能量的图像质量。结论：宝石CT能谱扫描能显著减少受检部位的金属伪影与硬化伪影,使含金属植入物的受检部位CT图像质量明显提高,具有较高的临床价值。     

锥形束CT图像分割方法研究进展

图像引导放射治疗(IGRT)是一种可视化的影像引导放疗技术, 具有提高肿瘤靶区剂量, 降低正常器官受照剂量等诸多优点。锥形束CT(CBCT)是IGRT中最常用的医学图像之一, 对CBCT进行快速、准确的靶区及危及器官的分割对放疗具有重大意义。目前的研究方法主要有基于配准的分割方法和基于深度学习的分割方法。本研究针对CBCT图像分割方法、存在问题及发展方向进行综述。     

基于深度学习的计算机辅助诊断系统在肺癌早期诊断中的应用与进展

胸部CT扫描是肺癌早期筛查和诊断的主要检查手段,应用于胸部影像诊断领域的基于深度学习的计算机辅助诊断(CAD)系统可对CT图像上的肺结节进行检测和分类。深度学习技术可提高CAD系统的性能,尤其是在提高肺结节检测的准确率和降低假阳性率方面。笔者就CAD系统中的深度学习模型在肺结节中的应用现状和研究进展作一综述。     

MRI-based treatment planning for brain stereotactic radiosurgery: Dosimetric validation of a learning-based pseudo-CT generation method

Magnetic resonance imaging (MRI)-only radiotherapy treatment planning is attractive since MRI provides superior soft tissue contrast without ionizing radiation compared with computed tomography (CT). However, it requires the generation of pseudo CT from MRI images for patient setup and dose calculation. Our machine-learning-based method to generate pseudo CT images has been shown to provide pseudo CT images with excellent image quality, while its dose calculation accuracy remains an open question. In this study, we aim to investigate the accuracy of dose calculation in brain frameless stereotactic radiosurgery (SRS) using pseudo CT images which are generated from MRI images using the machine learning-based method developed by our group. We retrospectively investigated a total of 19 treatment plans from 14 patients, each of whom has CT simulation and MRI images acquired during pretreatment. The dose distributions of the same treatment plans were calculated on original CT simulation images as ground truth, as well as on pseudo CT images generated from MRI images. Clinically-relevant DVH metrics and gamma analysis were extracted from both ground truth and pseudo CT results for comparison and evaluation. The side-by-side comparisons on image quality and dose distributions demonstrated very good agreement of image contrast and calculated dose between pseudo CT and original CT. The average differences in Dose-volume histogram (DVH) metrics for Planning target volume (PTVs) were less than 0.6%, and no differences in those for organs at risk at a significance level of 0.05. The average pass rate of gamma analysis was 99%. These quantitative results strongly indicate that the pseudo CT images created from MRI images using our proposed machine learning method are accurate enough to replace current CT simulation images for dose calculation in brain SRS treatment. This study also demonstrates the great potential for MRI to completely replace CT scans in the process of simulation and treatment planning.     

The use of deep learning towards dose optimization in low-dose computed tomography: A scoping review

IntroductionLow-dose computed tomography tends to produce lower image quality than normal dose computed tomography (CT) although it can help to reduce radiation hazards of CT scanning. Research has shown that Artificial Intelligence (AI) technologies, especially deep learning can help enhance the image quality of low-dose CT by denoising images. This scoping review aims to create an overview on how AI technologies, especially deep learning, can be used in dose optimisation for low-dose CT.MethodsLiterature searches of ProQuest, PubMed, Cinahl, ScienceDirect, EbscoHost Ebook Collection and Ovid were carried out to find research articles published between the years 2015 and 2020. In addition, manual search was conducted in SweMed+, SwePub, NORA, Taylor & Francis Online and Medic.ResultsFollowing a systematic search process, the review comprised of 16 articles. Articles were organised according to the effects of the deep learning networks, e.g. image noise reduction, image restoration. Deep learning can be used in multiple ways to facilitate dose optimisation in low-dose CT. Most articles discuss image noise reduction in low-dose CT.ConclusionDeep learning can be used in the optimisation of patients’ radiation dose. Nevertheless, the image quality is normally lower in low-dose CT (LDCT) than in regular-dose CT scans because of smaller radiation doses. With the help of deep learning, the image quality can be improved to equate the regular-dose computed tomography image quality.Implications to practiceLower dose may decrease patients’ radiation risk but may affect the image quality of CT scans. Artificial intelligence technologies can be used to improve image quality in low-dose CT scans. Radiologists and radiographers should have proper education and knowledge about the techniques used.     

基于U-Net模型从低能锥形束CT图像实现高能成像的研究

目的基于U-Net网络深度学习的方法, 实现在放疗临床中低能锥形束CT(CBCT)图像转换成高能CBCT图像, 以期提供双能CBCT成像图像基础且降低辐射剂量。方法利用放疗机载CBCT设备采集CIRS电子密度模体和CIRS头部体模在80和140 kV能量下的CBCT图像数据, 数据集按10∶1分为训练集和测试集。利用U-Net网络从低能量(80 kV)CBCT图像预测高能量(140 kV)下CBCT图像。采用平均绝对误差(MAE)、结构相似度指数(SSIM)、信噪比(SNR)和峰值信号噪声比(PSNR)4种度量指标, 定量评价预测高能CBCT图像。结果预测高能图像与真实高能图像之间总体结构差异较小(SSIM:0.993 ±0.003)。预测高能图像噪声较低(SNR:15.33±4.06), 但组织间分辨力有损失。预测高能图像比真实高能图像平均CT值偏低, 在低密度组织中差异较小(<10 HU, P > 0.05), 而在高密度组织中差异大(< 21 HU, t = -7.92, P < 0.05)。结论利用深度学习方法可以从低能CBCT图像获得结构相似度高的高能...     

基于深度学习的癌症计算机辅助分类诊断研究进展

近年来,基于深度学习的医学影像分类研究迅猛发展,已成为癌症计算机辅助诊断(CAD)领域的研究热点。目前,基于公开数据库(如LIDC、INbreast、DDSM公共数据库)进行的研究中,对肺结节良恶性分类研究、乳腺病灶的良恶性鉴别诊断的研究报道较为全面。对基于深度学习的肺CT图像、乳腺X线摄影图像的癌症计算机辅助分类诊断研究进展予以综述。     

Application of Artificial Intelligence to Cardiovascular Computed Tomography

Cardiovascular computed tomography (CT) is among the most active fields with ongoing technical innovation related to image acquisition and analysis. Artificial intelligence can be incorporated into various clinical applications of cardiovascular CT, including imaging of the heart valves and coronary arteries, as well as imaging to evaluate myocardial function and congenital heart disease. This review summarizes the latest research on the application of deep learning to cardiovascular CT. The areas covered range from image quality improvement to automatic analysis of CT images, including methods such as calcium scoring, image segmentation, and coronary artery evaluation.     

基于个人电脑的肝脏CT灌注软件开发

目的开发能够运行在个人电脑上的肝脏CT灌注软件，计算肝脏动脉、门脉灌注参数，并绘制相应的伪彩色灌注图。方法在个人电脑上，使用Delphi 7．0软件开发平台开发包括医学数字成像和通信标准(DICOM)文件读取、存储、显示接口的平台，在该平台之上开发肝脏灌注后处理软件PerfX。结果PeffX基于Windows2000／XP操作系统，能够在CPU 500MHz以上，内存128M以上的个人电脑上流畅地运行，可以测量特定感兴趣区的动脉、门脉灌注和肝灌注指数，并可以绘制相应的伪彩色灌注图，在1幅图像中显示肝脏解剖和功能的结合图像。结论PerfX在个人电脑上实现了从DICOM文件底层处理到肝脏CT灌注算法和灌注图绘制的后处理过程。     

Automated prediction of dosimetric eligibility of patients with prostate cancer undergoing intensity-modulated radiation therapy using a convolutional neural network

The quality of radiotherapy has greatly improved due to the high precision achieved by intensity-modulated radiation therapy (IMRT). Studies have been conducted to increase the quality of planning and reduce the costs associated with planning through automated planning method; however, few studies have used the deep learning method for optimization of planning. The purpose of this study was to propose an automated method based on a convolutional neural network (CNN) for predicting the dosimetric eligibility of patients with prostate cancer undergoing IMRT. Sixty patients with prostate cancer who underwent IMRT were included in the study. Treatment strategy involved division of the patients into two groups, namely, meeting all dose constraints and not meeting all dose constraints, by experienced medical physicists. We used AlexNet (i.e., one of common CNN architectures) for CNN-based methods to predict the two groups. An AlexNet CNN pre-trained on ImageNet was fine-tuned. Two dataset formats were used as input data: planning computed tomography (CT) images and structure labels. Five-fold cross-validation was used, and performance metrics included sensitivity, specificity, and prediction accuracy. Class activation mapping was used to visualize the internal representation learned by the CNN. Prediction accuracies of the model with the planning CT image dataset and that with the structure label dataset were 56.7?±?9.7% and 70.0?±?11.3%, respectively. Moreover, the model with structure labels focused on areas associated with dose constraints. These results revealed the potential applicability of deep learning to the treatment planning of patients with prostate cancer undergoing IMRT.