基于STL文件的个性化外固定支具模型生成算法研究

目的 研究基于STL文件的个性化外固定支具生成算法,实现通过3D扫描、模型生成和3D打印加工个性化外固定支具。与传统石膏固定相比,该方法获得的外固定支具具有与肢体外表面贴合性好、透气性好、重量轻的优点。 方法 为生成个性化外固定支具模型,对模型生成的关键算法进行了研究。用3D扫描仪扫描肢体获得点云文件并将其转化为STL格式文件;对三角形面片做切割、偏移、三角化和镂空等处理,从而生成透气性好并与肢体外表面完全贴合的外固定支具模型;最后利用3D打印加工外固定支具。 结果 使用改进后的点偏移算法,解决了点偏移在平面和尖角处严重变形的问题。使用正切值排序法对轮廓环上的点进行顺时针排序,获得三角化截面轮廓。使用典型立体“刀具”数学模型与STL文件空间相交方法实现了对模型的镂空加工。 结论 以手腕为例,对改进后的模型生成算法进行了验证,获得了与肢体表面贴合度很高的个性化外固定支具模型,证明该模型生成算法有效且实用。     

基于内耳体素模型的膜迷路三维可视化

目的探讨基于内耳体素模型三维可视化膜迷路。方法磁共振显微成像颞骨扫描影像数据使用3D Slicer软件进行表面模型体裁剪,将内耳体素模型通过表面绘制和体绘制混合成像三维显示膜迷路。结果通过内耳体素模型可以三维可视化膜迷路,且内耳体素模型文件较表面模型文件容量更小。结论内耳体素模型对于内耳解剖学习和研究有重要意义。     

双侧内耳统计形状模型

目的 探讨建立双侧内耳统计形状模型的方法和影响因素,并探讨使用平均双侧内耳模型作为带空间信息的标准模型。 方法 通过Otsu法以及手动分割37例双侧内耳模型,对模型进行配准并生成统计形状模型,观察测量标准模型空间信息。 结果 Otsu 法可以分割获取高质量内耳模型,模型配准后使用Statismo 0.81软件可以生成稳定的内耳统计形状模型,其导出的平均模型空间方向信息具有代表性。结论 双侧内耳统计形状模型可以导出平均模型作为带空间方向信息的标准模型。     

3D打印个性化康复矫形器的设计制作

目的:基于3D打印技术个性化康复矫形器的数字化设计制作研究,提出一种3D打印制作康复矫形器的可行性方案。方法:(1)用MRI的T1-mDIXON-W扫描模式扫描上肢,采集上肢肘以下前臂部分DICOM格式影像数据刻录成光盘,将光盘的数据导入Mimics软件重建出皮肤轮廓模型并导出STL格式文件,将STL格式文件导入Geomagic studio软件进行表面优化、网格划分、精确曲面等操作实体模型后保存为STP格式文件,将STP格式模型文件导入到Abaqus软件进行有限元结构的优化分析,设置10组不同的优化数值(5%、10%、15%、20%、25%、30%、35%、40%、45%、50%)分别进行拓扑优化,从而找到合适的优化数值,实现模型结构的优化,根据优化结果图和应力云图,指导模型的挖孔设计,设计好矫形器的模型,最后用尼龙材料通过3D打印机打印出来;(2)通过30个志愿者[男15人,女15人,年龄(24.26±2.38)岁,体质量(59.67±10.63)kg)]佩戴该方案设计制作的3D打印个性化矫形器和传统方法手工制作的矫形器,填写魁北克辅助科技满意度评估调查表,将表格数据用SPSS 19.0软件进行统计学分析。结果:(1)通过有限元拓扑优化分析,经过多次尝试,发现优化的体积参数设置在10%~20%之间,模型的生物力学性能影响较小,对照有限元优化结果指导模型的挖孔设计;(2)分析30个志愿者填写的魁北克辅助科技使用者满意度评估量表,可见矫形器使用的舒适度、矫形器使用的简易度、矫形器的性价比、矫形器的重量、外表美观性、矫形器的透气性均呈现出显著性(P0.05),意味着矫形器这6项指标有着统计意义上的差异,得出3D打印的矫形器在透气性和重量等方面优于传统矫形器。结论:本研究开发的3D打印康复矫形器的设计制作方案是可用的,能够为将来3D打印康复矫形器的大量临床应用提供参考。     

基于虚拟中国人数据集的鼻部及颞骨解剖结构三维重建

目的：为可视化虚拟人体模型鼻部及颞骨解剖结构的三维重建探索一种可行的方法。方法：利用3D-Slicer软件进行鼻部及颞骨部分解剖结构的三维重建。对单层图片进行图象分割及提取，处理后的体数据导入3D—Slicer，选择阈值进行进一步的图像分割，产生感兴趣区的标志图，进而重建出组织结构的三维表面模型。结果：成功重建了四组鼻窦，鼻中隔，中下鼻甲，颞骨，鼓室，乳突气房，乙状窦，颈内动脉的三维表面模型，并可显示不同结构间的毗邻关系与空间定位。结论：基于中国第一号虚拟人数据集，用3—D软件可以实现鼻部颞骨部分解剖结构三维可视化，便于对该部解剖结构的观察和理解。     

M3D:面向3D打印的医学图像重建平台

本研究旨在构建面向3D打印的医学图像3D重建平台,实现医学影像数据的目标分割、3D重建与3D打印所需区域裁剪与输出功能。该软件平台采用Qt设计用户界面,调用ITK、VTK类库处理医学影像数据。软件平台实现了医学影像数据的输入输出及处理、分割与3D重建等功能;在3D打印模型输出前,可实现对感兴趣区域的截取。以肘关节为例,利用该平台可实现模型重建,截取区域输出STL文件并完成3D打印。结果表明,该软件平台操作简单、界面简洁,可以从医学影像数据中重建得到3D模型,并进行3D打印,具有重要的临床应用价值。     

基于手持式3D扫描仪的人体解剖学实物标本三维数字化应用

目的 将珍贵的人体实物标本数字化为3D实物标本，使其可自由分享和随意编辑而不受网络、版权、专业软件等限制，直接插入教学PPT中应用于课堂，改进教学方式；或者插入Word中制作3D数字化人体解剖实物图谱。方法 用手持式三维扫描仪以及搭配的工作软件扫描实物标本，经过软件内置的智能模式生成3D数字化标本，保存并导出为.obj格式的3D文件。再经过Windows10系统自带的软件进行涂色、标注等加工后保存为.3mf格式的3D文件，插入到2019版PPT和Word中展示和观察。结果 数字化后的3D标本与实物标本形态结构非常接近，应用于课堂PPT演示和Word文件观察清晰流畅，3D原文件、PPT和Word文件均可自由拷贝和分享。结论 珍贵的人体实物标本可快速而高质量地数字化为3D标本，起到永久保存的作用，具有用途广，限制少等优点，并补充商业产品的不足。     

基于条件生成对抗网络的三维肝脏及肿瘤区域自动分割

肝脏计算机断层扫描成像(CT)的三维(3D)肝脏和肿瘤分割对于辅助医生的诊断及预后具有非常重要的临床价值。为了准确快速地分割肝脏及肿瘤区域,本文提出了一种基于条件生成对抗网络(cGAN)的肿瘤3D条件生成对抗分割网络(T3scGAN),同时采用了一个由粗到细的3D自动分割框架对肝脏及肿瘤区域实施精准分割。本文采用2017年肝脏和肿瘤分割挑战赛(LiTS)公开数据集中的130个病例进行训练、验证和测试T3scGAN模型。最终3D肝脏区域分割的验证集和测试集的平均戴斯(Dice)系数分别为0.963和0.961,而3D肿瘤区域分割的验证集和测试集的平均Dice系数分别为0.819和0.796。实验结果表明,提出的T3scGAN模型能够有效地分割3D肝脏及其肿瘤区域,因此能够更好地辅助医生进行肝脏肿瘤的精准诊断和治疗。     

基于3D U-Net 实现人体耳软骨MRI图像的解剖结构分割

自体肋软骨雕刻法是目前治疗先天性小儿畸形的临床标准疗法,而耳软骨组织工程和3D生物打印是有前景的治疗方案。可是,这些治疗方案的核心—(复合物)支架构造缺乏基于医学图像的耳软骨自动分割方法。基于3D U-Net提出改进的网络模型,能够实现MRI图像的人体耳软骨解剖结构的自动分割。该网络模型结合残差结构和多尺度融合等设计,在减少网络参数量的同时实现12个耳软骨解剖结构的精确分割。首先,使用超短回波时间(UTE)序列采集40名志愿者单侧外耳的MRI图像;然后,对所采集的图像进行预处理、耳软骨和多解剖结构手动标注;接下来,划分数据集训练改进的3D U-Net模型,其中32例数据作为训练集、4例为验证集、4例为测试集;最后,使用三维全连接条件随机场对网络输出结果进行后处理。模型经过10折交叉验证后,耳软骨12个解剖结构的自动分割结果的平均Dice相似度系数(DSC)和平均95%豪斯多夫距离(HD95)分别为0.818和1.917,相比于使用基础的3D U-Net模型,DSC指标分别提高6.0%,HD95指标降低了3.186,其中耳软骨关键结构耳轮和对耳轮的DSC指标达到了0.907和0.901。实验结果表明,所提出的深度学习方法与专家手动标注两者之间的结果非常接近。在临床应用中,根据患者健侧UTE核磁图像,本研究提出的方法既可以为现有自体肋软骨雕刻法快速、自动生成三维个性化雕刻模板,也可以为组织工程或者3D生物打印技术构建耳软骨复合物支架提供高质量的可打印模型。     

基于级联3D U-Net的CT和MR视交叉自动分割方法

目的:基于级联3D U-Net,利用配对患者头颈部数据［CT和磁共振图像（MRI）］,取得比仅CT数据更高分割精度的视交叉自动分割结果。方法:该级联3D U-Net由一个原始3D U-Net和改进的3D D-S U-Net（3D Deeply-Supervised U-Net）组成,实验使用了60例患者头颈部CT图像及MRI图像（T1和T2模态）,其中随机选取15例患者数据作为测试集,并使用相似性系数（DSC）评估视交叉的自动分割精度。结果:对于测试集中的所有病例,采用多模态数据（CT和MRI）的视交叉的DSC为0.645±0.085,采用单模态数据（CT）的视交叉的DSC为0.552±0.096。结论:基于级联3D U-Net的多模态自动分割模型能够较为准确地实现视交叉的自动分割,且优于仅利用单模态数据的方法,可以辅助医生提高放疗计划制定的工作效率。     

Software for browsing sectioned images of a dog body and generating a 3D model

The goals of this study were (1) to provide accessible and instructive browsing software for sectioned images and a portable document format (PDF) file that includes three‐dimensional (3D) models of an entire dog body and (2) to develop techniques for segmentation and 3D modeling that would enable an investigator to perform these tasks without the aid of a computer engineer. To achieve these goals, relatively important or large structures in the sectioned images were outlined to generate segmented images. The sectioned and segmented images were then packaged into browsing software. In this software, structures in the sectioned images are shown in detail and in real color. After 3D models were made from the segmented images, the 3D models were exported into a PDF file. In this format, the 3D models could be manipulated freely. The browsing software and PDF file are available for study by students, for lecture for teachers, and for training for clinicians. These files will be helpful for anatomical study by and clinical training of veterinary students and clinicians. Furthermore, these techniques will be useful for researchers who study two‐dimensional images and 3D models. Anat Rec, 299:81–87, 2016. © 2015 Wiley Periodicals, Inc.     

3D interactive model of lumbar spinal structures of anesthetic interest

A 3D model of lumbar structures of anesthetic interest was reconstructed from human magnetic resonance (MR) images and embedded in a Portable Document Format (PDF) file, which can be opened by freely available software and used offline. The MR images were analyzed using a specific 3D software platform for biomedical data. Models generated from manually delimited volumes of interest and selected MR images were exported to Virtual Reality Modeling Language format and were presented in a PDF document containing JavaScript‐based functions. The 3D file and the corresponding instructions and license files can be downloaded freely at http://diposit.ub.edu/dspace/handle/2445/44844?locale=en . The 3D PDF interactive file includes reconstructions of the L3–L5 vertebrae, intervertebral disks, ligaments, epidural and foraminal fat, dural sac and nerve root cuffs, sensory and motor nerve roots of the cauda equina, and anesthetic approaches (epidural medial, spinal paramedial, and selective nerve root paths); it also includes a predefined sequential educational presentation. Zoom, 360° rotation, selective visualization, and transparency graduation of each structure and clipping functions are available. Familiarization requires no specialized informatics knowledge. The ease with which the document can be used could make it valuable for anatomical and anesthetic teaching and demonstration of patient information. Clin. Anat. 28:205–212, 2015. © 2014 Wiley Periodicals, Inc.     

A detailed 3D model of the guinea pig cochlea

Several partial models of cochlear subparts are available. However, a complete 3D model of an intact cochlea based on actual histological sections has not been reported. Hence, the aim of this study was to develop a novel 3D model of the guinea pig cochlea and conduct post-processes on this reconstructed model. We used a combination of histochemical processing and the method of acquiring section data from the visible human project (VHP) to obtain a set of ideal raw images of cochlear sections. After semi-automatic registration and accurate manual segmentation with professional image processing software, one set of aligned data and six sets of segmented data were generated. Finally, the segmented structures were reconstructed by 3D Slicer (a professional imaging process and analysis tool). Further, post-processes including 3D visualization and a virtual endoscope were completed to improve visualization and simulate the course of the cochlear implant through the scala tympani. The 3D cochlea model contains the main six structures: (1) the inner wall, (2) modiolus and spiral lamina, (3) cochlea nerve and spiral ganglion, (4) spiral ligament and inferior wall of cochlear duct, (5) Reissner’s membrane and (6) tectorial membrane. Based on the results, we concluded that ideal raw images of cochlear sections can be acquired by combining the processes of conventional histochemistry and photographing while slicing. After several vital image processing and analysis steps, this could further generate a vivid 3D model of the intact cochlea complete with internal details. This novel 3D model has great potential in teaching, basic medical research and in several clinical applications. Electronic supplementary material The online version of this article (doi:) contains supplementary material, which is available to authorized users.     

Effects of CT image segmentation methods on the accuracy of long bone 3D reconstructions

An accurate and accessible image segmentation method is in high demand for generating 3D bone models from CT scan data, as such models are required in many areas of medical research. Even though numerous sophisticated segmentation methods have been published over the years, most of them are not readily available to the general research community. Therefore, this study aimed to quantify the accuracy of three popular image segmentation methods, two implementations of intensity thresholding and Canny edge detection, for generating 3D models of long bones. In order to reduce user dependent errors associated with visually selecting a threshold value, we present a new approach of selecting an appropriate threshold value based on the Canny filter. A mechanical contact scanner in conjunction with a microCT scanner was utilised to generate the reference models for validating the 3D bone models generated from CT data of five intact ovine hind limbs. When the overall accuracy of the bone model is considered, the three investigated segmentation methods generated comparable results with mean errors in the range of 0.18-0.24 mm. However, for the bone diaphysis, Canny edge detection and Canny filter based thresholding generated 3D models with a significantly higher accuracy compared to those generated through visually selected thresholds. This study demonstrates that 3D models with sub-voxel accuracy can be generated utilising relatively simple segmentation methods that are available to the general research community.     

A 2D driven 3D vessel segmentation algorithm for 3D digital subtraction angiography data

Cerebrovascular disease is among the leading causes of death in western industrial nations. 3D rotational angiography delivers indispensable information on vessel morphology and pathology. Physicians make use of this to analyze vessel geometry in detail, i.e. vessel diameters, location and size of aneurysms, to come up with a clinical decision. 3D segmentation is a crucial step in this pipeline. Although a lot of different methods are available nowadays, all of them lack a method to validate the results for the individual patient. Therefore, we propose a novel 2D digital subtraction angiography (DSA)-driven 3D vessel segmentation and validation framework. 2D DSA projections are clinically considered as gold standard when it comes to measurements of vessel diameter or the neck size of aneurysms. An ellipsoid vessel model is applied to deliver the initial 3D segmentation. To assess the accuracy of the 3D vessel segmentation, its forward projections are iteratively overlaid with the corresponding 2D DSA projections. Local vessel discrepancies are modeled by a global 2D/3D optimization function to adjust the 3D vessel segmentation toward the 2D vessel contours. Our framework has been evaluated on phantom data as well as on ten patient datasets. Three 2D DSA projections from varying viewing angles have been used for each dataset. The novel 2D driven 3D vessel segmentation approach shows superior results against state-of-the-art segmentations like region growing, i.e. an improvement of 7.2% points in precision and 5.8% points for the Dice coefficient. This method opens up future clinical applications requiring the greatest vessel accuracy, e.g. computational fluid dynamic modeling.     

Assessment of DICOM Viewers Capable of Loading Patient-specific 3D Models Obtained by Different Segmentation Platforms in the Operating Room

Patient-specific 3D models obtained by the segmentation of volumetric diagnostic images play an increasingly important role in surgical planning. Surgeons use the virtual models reconstructed through segmentation to plan challenging surgeries. Many solutions exist for the different anatomical districts and surgical interventions. The possibility to bring the 3D virtual reconstructions with native radiological images in the operating room is essential for fostering the use of intraoperative planning. To the best of our knowledge, current DICOM viewers are not able to simultaneously connect to the picture archiving and communication system (PACS) and import 3D models generated by external platforms to allow a straight integration in the operating room. A total of 26 DICOM viewers were evaluated: 22 open source and four commercial. Two DICOM viewers can connect to PACS and import segmentations achieved by other applications: Synapse 3D® by Fujifilm and OsiriX by University of Geneva. We developed a software network that converts diffuse visual tool kit (VTK) format 3D model segmentations, obtained by any software platform, to a DICOM format that can be displayed using OsiriX or Synapse 3D. Both OsiriX and Synapse 3D were suitable for our purposes and had comparable performance. Although Synapse 3D loads native images and segmentations faster, the main benefits of OsiriX are its user-friendly loading of elaborated images and it being both free of charge and open source.     

Fast prostate segmentation in 3D TRUS images based on continuity constraint using an autoregressive model

In this article a new slice-based 3D prostate segmentation method based on a continuity constraint, implemented as an autoregressive (AR) model is described. In order to decrease the propagated segmentation error produced by the slice-based 3D segmentation method, a continuity constraint was imposed in the prostate segmentation algorithm. A 3D ultrasound image was segmented using the slice-based segmentation method. Then, a cross-sectional profile of the resulting contours was obtained by intersecting the 2D segmented contours with a coronal plane passing through the midpoint of the manually identified rotational axis, which is considered to be the approximate center of the prostate. On the coronal cross-sectional plane, these intersections form a set of radial lines directed from the center of the prostate. The lengths of these radial lines were smoothed using an AR model. Slice-based 3D segmentations were performed in the clockwise and in the anticlockwise directions, where clockwise and anticlockwise are defined with respect to the propagation directions on the coronal view. This resulted in two different segmentations for each 2D slice. For each pair of unmatched segments, in which the distance between the contour generated clockwise and that generated anticlockwise was greater than 4 mm, a method was used to select the optimal contour. Experiments performed using 3D prostate ultrasound images of nine patients demonstrated that the proposed method produced accurate 3D prostate boundaries without manual editing. The average distance between the proposed method and manual segmentation was 1.29 mm. The average intraobserver coefficient of variation (i.e., the standard deviation divided by the average volume) of the boundaries segmented by the proposed method was 1.6%. The average segmentation time of a 352 x 379 x 704 image on a Pentium IV 2.8 GHz PC was 10 s.     

A software tool for 2D/3D visualization and analysis of phase-space data generated by Monte Carlo modelling of medical linear accelerators

A computer program has been developed for novel 2D/3D visualization and analysis of the phase-space parameters of Monte Carlo simulations of medical accelerator radiation beams. The software is written in the IDL language and reads the phase-space data generated in the BEAMnrc/BEAM Monte Carlo code format. Contour and colour-wash plots of the fluence, mean energy, energy fluence, mean angle, spectra distribution, energy fluence distribution, angular distribution, and slices and projections of the 3D ZLAST distribution can be calculated and displayed. Based on our experience of using it at Massachusetts General Hospital, the software has proven to be a useful tool for analysis and verification of the Monte Carlo generated phase-space files. The software is in the public domain.