Automatic segmentation of thoracic CT images using three deep learning models

PurposeDeep learning (DL) techniques are widely used in medical imaging and in particular for segmentation. Indeed, manual segmentation of organs at risk (OARs) is time-consuming and suffers from inter- and intra-observer segmentation variability. Image segmentation using DL has given very promising results. In this work, we present and compare the results of segmentation of OARs and a clinical target volume (CTV) in thoracic CT images using three DL models.Materials and methodsWe used CT images of 52 patients with breast cancer from a public dataset. Automatic segmentation of the lungs, the heart and a CTV was performed using three models based on the U-Net architecture. Three metrics were used to quantify and compare the segmentation results obtained with these models: the Dice similarity coefficient (DSC), the Jaccard coefficient (J) and the Hausdorff distance (HD).ResultsThe obtained values of DSC, J and HD were presented for each segmented organ and for the three models. Examples of automatic segmentation were presented and compared to the corresponding ground truth delineations. Our values were also compared to recent results obtained by other authors.ConclusionThe performance of three DL models was evaluated for the delineation of the lungs, the heart and a CTV. This study showed clearly that these 2D models based on the U-Net architecture can be used to delineate organs in CT images with a good performance compared to other models. Generally, the three models present similar performances. Using a dataset with more CT images, the three models should give better results.     

Update on sports imaging

Sports Imaging has dramatically increased in the past decade with increasing number of adolescents, young and middle-aged adults participating in non-competitive/hobby sports. Therefore, sports injuries are no longer confined to elite athletes. Furthermore, newer forms of sports such as mountain climbing, pickle ball and curling etc. are gaining popularity. Majority of the injuries in sports medicine are from musculoskeletal trauma. Therefore, it is imperative that the musculoskeletal radiologist becomes familiar with various sports related injury patterns as these are commonly encountered in daily practice. This update aims to briefly encapsulate the major aspects of sports imaging. It includes the imaging manifestations of various types of musculoskeletal injuries on different modalities (commonly US and MRI) and briefly mentions the various image guided interventions, performed both on the sports field and in the hospital setting.     

Organ-based tube current modulation in chest CT. A comparison of three vendors

IntroductionOrgan-based tube current modulation (OBTCM) is designed for anterior dose reduction in Computed Tomography (CT). The purpose was to assess dose reduction capability in chest CT using three organ dose modulation systems at different kVp settings. Furthermore, noise, diagnostic image quality and tumour detection was assessed.MethodsA Lungman phantom was scanned with and without OBTCM at 80–135/140 kVp using three CT scanners; Canon Aquillion Prime, GE Revolution CT and Siemens Somatom Flash. Thermo-luminescent dosimeters were attached to the phantom surface and all scans were repeated five times. Image noise was measured in three ROIs at the level of the carina. Three observers visually scored the images using a fivestep scale. A Wilcoxon Signed-Rank test was used for statistical analysis of differences.ResultsUsing the GE revolution CT scanner, dose reductions between 1.10 mSv (12%) and 1.56 mSv (24%) (p < 0.01) were found in the anterior segment and no differences posteriorly and laterally. Total dose reductions between 0.64 (8%) and 0.91 mSv (13%) were found across kVp levels (p < 0.00001). Maximum noise increase with OBTCM was 0.8 HU. With the Canon system, anterior dose reductions of 6–10% and total dose reduction of 0.74–0.76 mSv across kVp levels (p < 0.001) were found with a maximum noise increase of 1.1 HU. For the Siemens system, dose increased by 22–51% anteriorly; except at 100 kVp where no dose difference was found. Noise decreased by 1 to 1.5 HU.ConclusionOrgan based tube current modulation is capable of anterior and total dose reduction with minimal loss of image quality in vendors that do not increase posterior dose.Implications for practiceThis research highlights the importance of being familiar with dose reduction technologies.     

基于自动勾画和快速蒙特卡罗计算的放射治疗全身器官剂量数据库平台的可行性研究

目的:探讨建立一种放射治疗全身器官剂量数据库平台的可行性。方法:使用基于深度学习的自动勾画软件DeepViewer?1例食管癌患者的全身CT上勾画全身器官，然后利用基于GPU加速的蒙特卡罗软件ARCHER计算相应的器官剂量分布，最后利用Lyman-Kutcher-Burman（LKB）模型评估放疗患者正常组织并发症概率（NTCP）。结果:针对该病例，成功建立基于DeepViewer?ARCHER和LKB模型的全身器官剂量数据库，发现距离靶区越近的器官剂量越大，其中心脏与靶区间距离最小，剂量为14.11 Gy，但因其模型参数特殊，通过LKB模型计算的NTCP为0.00%；左、右肺的剂量分别为3.19和1.16 Gy，但是NTCP值却很大，分别为2.13%和1.60%。对于距离靶区较远的头颈部器官（视交叉、视神经和眼）和腹部器官（直肠、膀胱和股骨头）剂量分别约为9和2 mGy，并且NTCP均近似为0.00%。结论:研究结果证明通过自动勾画软件DeepViewer?蒙特卡罗软件ARCHER和LKB模型建立全身器官剂量数据库的可行性。     

肝脏CT动态增强扫描非刚性对位减影技术的质量评价

目的利用血管减影前后的吻合程度，探讨非刚性对位减影技术在肝脏CT动态增强检查中的质量评价。 方法收集2018年12月至2019年2月于中山大学附属第一医院行肝脏CT动态增强检查的50例患者资料，使用非刚性对位减影技术获得减影图像，对减影前后的图像质量进行评价，定性评价包括图像伪影情况及解剖错配程度，定量评价测量各血管（腹主动脉、肝右动脉、肝左动脉、门静脉主干、门静脉右支、门静脉左支）减影前后在动脉晚期、门静脉晚期的CT值。采用混合线性效应模型及拟合方程评估减影前后各血管的CT值曲线拟合程度。 结果减影后图像均无明显解剖结构错配导致的伪影，3名测量者一致性较高（ICC=0.844，P<0.001）。减影技术一定程度上会降低各血管的CT值，与强化分期不存在交互作用。无论是在动脉晚期还是门静脉晚期，减影前后各血管CT值走势一致，CT值变化曲线吻合。 结论非刚性对位减影技术可轻松实现肝脏CT动态增强检查的图像减影，并获得良好的图像质量。     

MRI图像融合技术在肛瘘评估中的初步应用

目的采用图像融合技术获得T2WI与T2WI-FS的融合图像，评估其在肛瘘及肛周结构显示中的优势。 方法2016年6月至2018年6月，前瞻性选择中山大学附属第一医院29例肛瘘患者进行肛管磁共振（MR）检查，采用图像融合技术获取T2WI与T2WI-FS的融合图像T2WI-Fusion，利用Fisher score算法计算瘘管及肛门括约肌的组织间分辨力Fisher值、脂肪与肛门括约肌间的Fisher值，评估融合图像中瘘管及肛周结构的显示情况。采用改进的双刺激连续质量量表（DSCQS）对T2WI-FS、T2WI、增强3D-VIBE和T2WI-Fusion序列图像进行主观图像质量评价。 结果29例患者均成功获得T2WI与T2WI-FS的融合图像T2WI-Fusion。T2WI-Fusion、T2WI瘘管与括约肌间Fisher均值分别为6.46、3.31，T2WI-Fusion图像对瘘管的显示优于T2WI序列图像（P<0.001）。T2WI-Fusion、T2WI-FS脂肪与括约肌间Fisher均值分别为10.61、2.45，T2WI-Fusion图像对括约肌的显示优于T2WI-FS序列图像（P<0.001）。T2WI-Fusion对瘘管与括约肌的图像质量评价总评分均高于T2WI-FS、T2WI、增强3D-VIBE序列（P<0.001）。 结论MRI图像融合技术同时具备T2WI及T2WI-FS的优势，无需增加扫描序列及扫描时间，且操作简单，花费时间短，显著提高病变及肛周解剖结构的对比度和图像质量。     

舌象分割技术的文献计量学分析

[目的]了解舌象分割技术的研究现况及发展趋势,为舌象客观化提供参考。[方法]基于中国知网、维普、万方及Pubmed数据库检索建库至2018年公开发表的舌象分割技术文献,利用Note Express 3.20、Excel 2016、SATI3.2、UCINET 6.0软件进行文献计量学及关键词共现网络分析。[结果]舌象分割技术文献研究整体呈上升趋势,北京工业大学居首位。主要基金资助为国家自然科学基金。根据普赖斯定律的要求,基本形成"核心作者群"。舌象分割以Snake模型和基于阈值的分割技术为主。关键词集中分布在"舌诊""图像分割""舌诊客观化"等方面。[结论]舌象分割技术研究仍需受到重视,国家政府应加大资助力度,加强研究机构间合作,尚存在一些舌象分割评价标准问题,需待改进。     

基于机器学习的脑胶质瘤多模态影像分析

脑部胶质瘤是临床中常见的一种原发性脑肿瘤，具有复发率高、死亡率高以及治愈率低的特点。常规临床诊断主要依靠计算机断层扫描(CT)和磁共振成像(MRI)检查技术进行鉴别。随着成像技术和机器学习方法的不断发展，多模态影像智能分析技术已经逐步成为研究热点，在脑胶质瘤的病灶分割测量、肿瘤分级、预后生存周期预测和基因型辨别等方面具有重要的应用前景。本文重点介绍基于机器学习和多模态影像在脑胶质瘤临床辅助诊断和预后评估中的应用进展。