Free form deformation and symmetry constraint‐based multimodal brain image registration using generative adversarial nets

Multi‐modal brain image registration has been widely applied to functional localisation,neurosurgery and computational anatomy.The existing registration methods based on the dense deformation fields involve too many parameters,which is not conducive to the exploration of correct spatial correspondence between the float and reference images.Meanwhile,the unidirectional registration may involve the deformation folding,which will result in the change of topology during registration.To address these issues,this work has presented an unsupervised image registration method using the free form deformation(FFD)and the symmetry constraint‐based generative adversarial networks(FSGAN).The FSGAN utilises the principle component analysis network‐based structural representations of the reference and float images as the inputs and uses the generator to learn the FFD model parameters,thereby producing two deformation fields.Meanwhile,the FSGAN uses two discriminators to decide whether the bilateral registration have been realised simultaneously.Besides,the symmetry constraint is utilised to construct the loss function,thereby avoiding the deformation folding.Experiments on BrainWeb,high grade gliomas,IXI and LPBA40 show that compared with state‐of‐the‐art methods,the FSGAN provides superior performance in terms of visual comparisons and such quantitative indexes as dice value,target registration error and computational efficiency.

Multi-modality liver image registration based on multilevel B-splines free-form deformation and L-BFGS optimal algorithm

A new coarse-to-fine strategy was proposed for nonrigid registration of computed tomography(CT) and magnetic resonance(MR) images of a liver.This hierarchical framework consisted of an affine transformation and a B-splines free-form deformation(FFD).The affine transformation performed a rough registration targeting the mismatch between the CT and MR images.The B-splines FFD transformation performed a finer registration by correcting local motion deformation.In the registration algorithm,the normalized mutual information(NMI) was used as similarity measure,and the limited memory Broyden-Fletcher- Goldfarb-Shannon(L-BFGS) optimization method was applied for optimization process.The algorithm was applied to the fully automated registration of liver CT and MR images in three subjects.The results demonstrate that the proposed method not only significantly improves the registration accuracy but also reduces the running time,which is effective and efficient for nonrigid registration.

Real-time accurate Free-Form Deformation in terms of triangular Bézier surfaces

We implemented accurate FFD in terms of triangular Bezier surfaces as matrix multiplications in CUDA and rendered them via OpenGL. Experimental results show that the proposed algorithm is more efficient than the previous GPU acceleration algorithm and tessel- lation shader algorithms.

The Skin Deformation of a 3D Virtual Human

This paper presents a skin deformation algorithm for creating 3D characters or virtual human models. The algorithm can be applied to rigid deformation, joint dependent localized deformation, skeleton driven deformation, cross contour deformation, and free-form deformation （FFD）. These deformations are computed and demonstrated with examples and the algorithm is applied to overcome the difficulties in mechanically simulating the motion of the human body by club-shape models. The techniques described in this article enables the reconstruction of dynamic human models that can be used in defining and representing the geometrical and kinematical characteristics of human motion.

Image Deformation by User-Defined Force Fields

We present a new method for image deformation. The warping technique provides smooth distortion with intuitive and easy manipulation. Driven by a restrained force field, the input image is deformed gradually and continuously. The method allows us to customize the force fields and the region of interest manually through some simple steps. Experimental results demonstrate the effectiveness and convenience of the approach.

个性化头部生物力学模型的开发及验证

本文提出了一种个性化头部生物力学模型建模方法,并对所生成模型进行了有效性验证。该方法以图斯特50th百分位头部生物力学模型为基础,基于目标模型头部CT数据,采用三维点云配准和自由网格变形法,实现具有详实脑组织结构的个性化头部生物力学模型的快速建模。通过重构经典尸体试验,发现该方法构建的个性化头部生物力学模型在运动学和生物力学响应方面与尸体试验结果高度一致,且与通过逆向工程方法开发的头部生物力学模型并无明显差异,验证了本文方法所开发模型的有效性。因此,本文方法可用于快速构建具有详细解剖学结构的个性化头部生物力学模型,为损伤生物力学、临床医学和法医鉴定等领域的数字化需求提供基础计算分析工具。

双作用液环压缩机壳体型线响应面分析及多目标优化

为探究双作用液环压缩机壳体型线对其内流场及性能的影响机理,采用DFFD参数化方法对多级双作用液环压缩机椭圆形壳体型线进行响应面优化分析,并结合NSGA-II多目标优化方法对吸排气区壳体型线进行协同优化。采用VOF多相流模型对该液环压缩机全流域进行非定常计算,双作用液环泵内相态场、压力场基本呈径向对称分布,吸排气区域及叶轮壳体内流场与单作用液环泵分布规律基本一致,随着级数增大液环内的气相区域逐渐减小。响应面结果表明吸气量的大小主要取决于吸气口始端和末端的壳体型线,当R_(A)值趋于0水平,R_(B)处于低水平时,吸气量趋于最大值。轴功率主要受吸、排气口始端位置壳体型线影响,R_(A),R_(C)值越小,轴功率越小。效率的大小主要取决于吸气口始端和末端的壳体型线,且R_(A)值越大,R_(B)值越小,效率越高。在额定出口压比条件下,以吸气量和效率作为优化目标对液环压缩机进行NSGA-II多目标优化,获得其Pareto前沿的最优效率设计J和和最优流量设计K,J方案流量和效率分别高于原方案1.40%和2.53%,K方案流量和效率分别高于原方案16.4%和0.61%。

数据驱动的两级轴流涡轮多自由度气动优化设计

针对涡轮气动优化过程中设计变量多、评估耗时长、传统完全替代精确函数评估的代理模型辅助算法全局搜索性较差等问题,采用基于预筛选策略的代理模型辅助差分进化算法(Pre-SADE),并结合直接操纵自由变形方法 (DFFD)实现多自由度参数化控制,建立了基于数据驱动的多级涡轮三维气动优化平台。以某两级轴流涡轮为研究对象,选取44个设计变量开展流道-叶片排联合气动优化设计。结果表明:优化后涡轮设计点等熵效率提高1.10%,流量增加2.16%,落压比降低0.94%,激波强度降低,径向二次流得到抑制,内部流动损失减小。此外,涡轮设计转速全工况特性均有所改善。该平台在保证优化效果的同时,可大幅减少设计变量和真实评估次数,有效适用于多级涡轮、多自由度气动优化问题。

Non-Rigid Medical Image Registration with Joint Histogram Estimation Based on Mutual Information

A mutual information-based non-rigid medical image registration algorithm is presented. An approximate function of Hanning windowed sinc is used as kernel function of partial volume (PV) interpolation to estimate the joint histogram, which is the key to calculating the mutual information. And a new method is proposed to compute the gradient of mutual information with respect to the model parameters. The transformation of object is modeled by a free-form deformation (FFD) based on B-splines. The experiments on 3D synthetic and real image data show that the algorithm can converge at the global optimum and restrain the emergency of local extreme.

基于EFFD参数化的风扇/压气机叶片-端壁一体化伴随优化设计

为解决传统扰动参数化方法的设计能力不足等问题,以拓展自由变形技术为基础开发相应参数化方法以改进伴随优化系统,并对典型跨声速风扇/压气机转子Rotor 67进行叶片-端壁一体化伴随优化。结果表明:经过伴随优化,Rotor 67转子在流量、压比等工况约束变化较小的前提下效率提升了0.74%,且整体特性同样得到了大幅改进,而优化前后的几何与流动变化表明,端区几何调整及叶片吸力面变化引起的吸力面加速减弱、激波强度降低、角区分离涡结构改进等,均是性能提升的内在原因。

基于自由变形技术的多投影颜色校正

为了解决多通道投影显示系统中各投影仪投影画面之间颜色不一致的问题,设计了一种基于自由变形技术的多投影颜色校正方法。首先,通过Bernstein基函数建立自由变形技术模型,建立各投影仪原始图像和摄像机拍摄图像之间的颜色转换关系;其次,采集原始图像集和摄像机拍摄的投影显示画面集,确定自由变形技术模型的参数;然后,通过Matlab分析验证投影图像各颜色通道之间是相互影响的;最后,对原始图像进行颜色扭曲,将本文颜色校正方法与传统方法进行直方图相似性评估对比。实验结果表明:对比广义颜色校正方法和基于B样条曲线的颜色校正方法,本文方法将投影画面的颜色强度平均差值在B通道减少了2.50,在G通道减少了2.34,在R通道减少了3.57,直方图的相关性提高了8.9%,巴氏距离降低了9.7%。基于自由变形技术的多投影颜色校正方法使投影画面衔接流畅,给用户带来更好的沉浸感受。

基于SST全湍流伴随的尾桨翼型优化方法

为解决当前翼型优化中广泛使用的冻结湍流黏性假设存在的固有缺陷和基于Spalart-Allmaras(S-A)全湍流伴随中湍流模型对气动力计算精度较差的问题,提出一套新的翼型优化方法,其耦合了全湍流连续伴随求解、剪切应力传递(SST)湍流模型封闭的雷诺平均Navier-Stokes(RANS)方程、自由变形参数化方法和动网格变形技术。基于所提方法,在气动力系数相较于S-A模型有更高捕捉精度的基础上,对NPL9615翼型以最大升阻比为优化目标,并与冻结湍流黏性假设方法对比。结果表明:所提方法将原有翼型的升阻比提高了16.39%,而冻结湍流黏性假设方法获得最终翼型的升阻比仅提高了原有翼型的9.84%,说明所提方法在最优外形的获取上要领先于冻结湍流黏性假设,并且当翼型周围的湍流动能显著提高时,其优势愈发扩大。

基于过度加速度失效模式的船型优化研究

选择第二代完整稳性中过度加速度的长期失效概率为优化目标函数,对船舶进行全船型线优化.采用自由变形技术作为船舶几何重构方法,优化方法采用可避免陷入局部陷阱的随机搜索全局优化算法.对某渔政船进行程序实验,优化后得出的新船型长期失效概率降低64.35%,有效降低了该船发生过度加速度的概率,验证了程序的实用性.

超大型组合翼伞开伞性能研究

为深入了解超大型组合翼伞与连续翼伞开伞过程的差异,文章通过数值模拟计算对超大型组合翼伞和连续翼伞的开伞性能进行了对比研究。首先,基于自由曲面变形(Free-Form-Deformation,FFD)理论建立了超大型翼伞的展向折叠模型,然后应用任意拉格朗日-欧拉(ArbitraryLagrange-Euler,ALE)方法对翼伞折叠模型进行了充气过程的流固耦合仿真计算。分析了超大型组合翼伞和连续翼伞充气过程中的伞衣外形、伞衣应力和气室充气规律,深入对比了超大型组合翼伞和连续翼伞的充满时间和开伞动载。研究表明,连续翼伞和组合翼伞的开伞性能差异很小,两者的充满时间相差不大,组合翼伞展开效果更好,动载之差在5%左右。与连续翼伞相比,组合翼伞更易于制造和使用。研究将为超大型翼伞的设计优化提供一定的参考。

基于CFD的飞翼布局无人机气动外形优化

飞翼布局因其独特的翼身融合的气动外形,大大提高了飞行器的有效升力面积,外形优化问题和布局优化对于此类构型气动性能的提升同样重要。本文为解决飞翼布局无人机气动外形优化问题,建立了高效的参数化建模方法,实现了适应复杂外形的几何参数化变形控制,将基于梯度的优化算法、离散伴随方法与基于RANS(Reynolds average Navier-Stokes)方程的计算流体力学(Computational fluid dynamics,CFD)方法相结合,对飞翼布局无人机完成了气动外形的优化减阻设计,升阻比提升了7.17%。优化结果表明,在满足约束要求的前提下,基于上述技术的气动优化设计方法对翼身融合类构型具有良好的适应性,能有效改善无人机的气动性能。

基于自由变形的航空叶片模型快速修正

使用传统重构方法对航空叶片模型进行重构需要对叶片实际点云进行点云分割,曲面拼接等复杂操作,重构效率较低。针对此问题,提出了一种基于自由变形的航空叶片模型快速修正方法。首先,将叶片设计模型离散点云与叶片实际点云数据进行点云配准;其次,通过KD-tree计算叶片模型的特征点集与实际点云数据的误差,结合自由变形算法实现叶片设计模型局部或整体误差快速修正;最后,进行拓扑重构,完成叶片设计模型变形修正。实验结果表明,与传统方法重构的叶片设计模型相比,所提方法算法效率与重构精度得到显著提升,将重构精度提升至0.092 8 mm,并且算法运算效率为传统重构方法的7.231倍。

An investigation on adaptively machining the leading and tailing edges of an SPF/DB titanium hollow blade using free-form deformation

Titanium hollow blades are characterized with lightweight and high structural strength, which are widely used in advanced aircraft engines nowadays. Superplastic forming/diffusion bonding （SPF/DB） combined with numerical control （NC） milling is a major solution for manufacturing titanium hollow blades. Due to the shape deviation caused by multiple heat and pressure cycles in the SPF/DB process, it is hard to manufacture the leading and tailing edges by the milling process. This paper presents a new adaptive machining approach using free-form deformation to solve this problem. The actual SPF/DB shape of a hollow blade was firstly inspected by an on-machine measurement method. The measured point data were matched to the nominal SPF/DB shape with an improved ICP algorithm afterwards, by which the point-pairs between the measurement points and their corresponding points on the nominal SPF/DB shape were established, and the maximum modification amount of the final nominal shape was constrained. Based on the displacements between the point-pairs, an accurate FFD volume was iteratively calculated. By embedding the final nominal shape in the deformation space, a new final shape of the hollow blade was built. Finally, a series of measurement and machining tests was performed, the results of which validated the feasibility of the proposed adaptive machining approach.

渔船减阻球鼻艏模型优化研究

为探究球鼻艏变化对渔船水动力性能的影响,在母型船直立艏方案的基础上,以总阻力为优化目标,进行增设球鼻艏设计.首先对渔船进行参数化建模,采用自由变形(FFD)方法对艏部的几何参数进行改造;然后使用拉丁超立方设计方法进行采样,将CFD数值模拟与BP神经网络相结合,建立以高度精确的数值模拟为优化特征的母型船预测模型;最后利用遗传算法对渔船进行目标优化,寻求总阻力优解的设计点.结果表明:基本船型配合增设优解的球鼻艏后,在3种航速情况下,总阻力分别减小0.30%、0.58%、1.02%.

Free-Form Deformation with Rational DMS-Spline Volumes

In this paper, we propose a novel free-form deformation （FFD） technique, RDMS-FFD （Rational DMS-FFD）, based on rational DMS-spline volumes. RDMS-FFD inherits some good properties of rational DMS-spline volumes and combines more deformation techniques than previous FFD methods in a consistent framework, such as local deformation, control lattice of arbitrary topology, smooth deformation, multiresolution deformation and direct manipulation of deformation. We first introduce the rational DMS-spline volume by directly generalizing the previous results related to DMS-splines. How to generate a tetrahedral domain that approximates the shape of the object to be deformed is also introduced in this paper. Unlike the traditional FFD techniques, we manipulate the vertices of the tetrahedral domain to achieve deformation results. Our system demonstrates that RDMS-FFD is powerful and intuitive in geometric modeling.