3D打印工艺下可降解植入物的降解速率进行预测，在医学领域的应用有重要的价值和意义。本文结合图像处理技术和3D打印技术，以医院提供的核磁共振图像为研究对象，进行如下方面的研究：

分析医学图像噪声类型，提出基于噪声检测的改进自适应双边滤波医学图像降噪算法。利用噪声的特性实现了图像噪声检测与标记，将椒盐噪声点替换为邻域内的非噪声点滤除图像中的椒盐噪声，再通过自适应双边滤波滤除图像高斯噪声，实现了保护图像边缘的图像混合噪声去除。降噪实验表明，本文降噪算法对混合噪声的滤除效果均优于传统降噪算法；边缘检测实验表明，该算法在保留边缘细节方面的能力优于传统算法。

对图片中的植入物区域进行分割，提出了基于Canny算子和距离正则化的图像分割算法。利用Canny算子对图像边缘分割精确且能够抑制噪音的优点及DRLSE空间连续演化的思想，结合Canny算子与DRLSE算法，设计了一种分割算法；将Canny算子的解析式代入DRLSE方程中替换DRLSE解析式中的边缘指示函数，推导出本文算法的解析式。图像分割实验表明，本文算法的分割精度与抗噪声能力均优于DRLSE算法。

通过DICOM文件的标签与图像分割结果计算植入物的体积；用一元线性回归模型结合机器学习方法实现对植入物降解速率的预测，将均方根误差作为损失函数指引优化方向，利用梯度下降法对一元线性回归模型进行优化，得到合适的一元线性预测模型。

开展3D打印可降解植入物降解速率预测试验。选择PCL作为可降解植入物原材料，熔融沉积成型3D打印机作为加工设备进行打印；由医院在保乳手术中将可降解假体植入患者体内，收集患者恢复期各阶段核磁共振影像数据；将本文提出的算法编写成为应用程序，测量出患者每一个恢复阶段植入物体积，把体积与植入时间制作成数据集并输入到一元线性回归预测模型中进行训练，得到预测模型；通过将植入物在患者体内降解的时间输入预测模型进行试验，对比预测值与真实值之间差距，获得本预测方法的精度。

The prediction of the degradation rate of biodegradable implants under 3D printing technology is of great value and significance in the medical field. Combined with image processing technology and 3D printing technology, this paper takes MRI images provided by hospitals as the research object and conducts the following researches:

Analyze the types of medical image noise, and propose an improved adaptive bilateral filtering medical image noise reduction algorithm based on noise detection. The characteristics of noise are used to realize image noise detection and labeling. The salt and pepper noise points are replaced with non-noise points in the neighborhood to filter out the salt and pepper noise in the image, and then the image Gaussian noise is filtered out through adaptive bilateral filtering to realize the protection of image edges. Image mixed noise removal. The noise reduction experiment shows that the noise reduction algorithm in this paper is better than the traditional noise reduction algorithm in filtering mixed noise; the edge detection experiment shows that the algorithm is better than the traditional algorithm in retaining edge details.

To segment the implant region in the picture, an image segmentation algorithm based on Canny operator and distance regularization is proposed. Using the advantages of Canny operator to segment the image edge accurately and suppressing noise and the idea of continuous evolution of DRLSE space, combining Canny operator and DRLSE algorithm, a segmentation algorithm is designed; the analytical formula of Canny operator is substituted into the DRLSE equation. The edge indicator function in the DRLSE analytic formula is derived from the analytic formula of the algorithm in this paper. Image segmentation experiments show that the segmentation accuracy and anti-noise ability of this algorithm are better than DRLSE algorithm.

Design methods for implant volume measurement and degradation rate prediction. Calculate the volume of the implant through the label of the DICOM file and the results of image segmentation; choose a one-element linear regression model combined with machine learning methods to predict the degradation rate of the implant, use the root mean square error as the loss function to guide the model optimization direction, and use the gradient The descent method optimizes the unary linear regression model to obtain a suitable unary linear prediction model.

The volume of implant was calculated by the label and image segmentation results of DICOM file. The unary linear regression model combined with machine learning method was used to predict the degradation rate of implants. The root mean square error was used as the loss function to guide the optimization direction. The gradient descent method was used to optimize the unary linear regression model, and the appropriate unary linear prediction model was obtained.

Carry out 3D printing degradable implant degradation rate prediction test. PCL is selected as the raw material for the degradable implant, and the fused deposition molding 3D printer is used as the processing equipment for printing; the hospital will implant the degradable prosthesis in the patient during the breast-conserving surgery, and collect the MRI data of the patient at each stage of the recovery period; The algorithm proposed in this paper is written into an application program to measure the implant volume at each recovery stage of the patient, make the volume and implantation time into a data set and input it into the unary linear regression prediction model for training to obtain the prediction model; The time of the degradation of the incoming substance in the patient's body is input to the prediction model for testing, and the difference between the predicted value and the true value is compared to obtain the accuracy of the prediction method.

[1] 余冬梅.3D打印:定制化医疗修复[J].金属世界,2016,(1):21-28. DOI:10.3969/j.issn.1000-6826.2016.01.06.

[2] 向声燚, 焦志伟, 苗剑飞,等. 聚己内酯3D打印成型工艺研究及力学性能分析[J]. 中国塑料, 2018, v.32;No.289(04):80-84.

[3] 陈琳.PCL的降解性能及其用于组织工程支架的制备方法研究[D].上海:东华大学,2007. DOI:10.7666/d.y1440959.

[4] Haryńska Agnieszka,Kucinska-Lipka Justyna,Sulowska Agnieszka,Gubanska Iga, Kostrzewa Marcin,Janik Helena. Medical-Grade PCL Based Polyurethane System for FDM 3D Printing-Characterization and Fabrication.[J]. Materials (Basel, Switzerland), 2019,12(6).

[5] 许园甫. 医学图像处理与分析的应用和意义[J]. 华夏医学, 2008, 21(6):1176-1177.

[6] 杨前华, 赵力. 一种基于小波高斯去噪声方法的迭代算法[J]. 通信技术, 2020(7):1641-1644.

[7] [1]CT图像的颅内血肿体积测量方法研究[D].安徽:安徽大学, 2019.

[8] Gonzalez R , Woods R , Eddins S . 数字图像处理(MATLAB版)[M]// 数字图像处理（MATLAB版）. 电子工业出版社, 2005.

[9] Noel G. Stoker,Nicholas J. Mankovich,Dan Valentino.Stereolithographic models for surgical planning: Preliminary report[J].口腔颌面外科杂志,1992,50(5).466-471.

[10] D'Urso PS,Arvier JF,Barker TM,et al.Stereolithographic biomodelling in cranio-maxillofacial surgery: a prospective trial.[J].Journal of Cranio-Maxillofacial Surgery,1999,27(1).

[11] Caldroney S J , Bourell L , Levine J , et al. Poster 78: CAD/CAM Virtual Surgery; A Comprehensive Review of Its Use in Various Applications in the Field of Oral and Maxillofacial Surgery[J]. Journal of Oral & Maxillofacial Surgery, 2012, 70(9):e89-e90.

[12] Saijo H , Igawa K , Kanno Y , et al. Maxillofacial reconstruction using custom-made artificial bones fabricated by inkjet printing technology[J]. Journal of Artificial Organs, 2009.

[13] Lawlor K T , Vanslambrouck J M , Higgins J W , et al. Cellular extrusion bioprinting improves kidney organoid reproducibility and conformation[J]. Nature Materials.

[14] 孙畅, 张超. 基于局部与非局部医学图像降噪技术的研究进展[J]. 中国医学物理学杂志, 2018, 035(005):565-572.

[15] Perona, P.,Institute of Electric and Electronic Engineer.Orientation diffusions[C]. //Computer Vision and Pattern Recognition, 1997. Proceedings., 1997 IEEE Computer Society Conference on.1997:710-716.

[16] Francine Catté, Lions P L , Morel J M , et al. Image Selective Smoothing and Edge Detection by Nonlinear Diffusion[J]. Siam Journal on Numerical Analysis, 1992, 29(1):845-866.

[17] Kumar B K S . Image denoising based on non-local means filter and its method noise thresholding[J]. Signal Image & Video Processing, 2013, 7(6):1211-1227.

[18] Dabov K ,Foi A ,Katkovnik V , et al. Image Denoising by Sparse 3-D Transform-Domain Collaborative Filtering[J]. IEEE Transactions on Image Processing, 2007, 16(8): 2080-2095.

[19] Qin J , Shen X , Mei F , et al. An Otsu multi-thresholds segmentation algorithm based on improved ACO[J]. The Journal of Supercomputing, 2019, 75(2):955-967.

[20] Liang Y , Fu J . Watershed Algorithm for Medical Image Segmentation Based on Morphology and Total Variation Model[J]. International Journal of Pattern Recognition and Artificial Intelligence, 2018, 33(5):1954019.1-1954019.10.

[21] 钟静, 黄文华. 3D打印技术辅助耳修复手术的研究进展[J]. 中国医学物理学杂志, 2016, 33(12):1286-1290.

[22] 孙涛,张鹤达,李建, 等.个案报道:基于3D打印技术的乳腺肿瘤精准手术治疗2例[J].中国肿瘤外科杂志,2016,8(4):235-239. DOI:10.3969/j.issn.1674-4136.2016.04.005.

[23] 张聚良,姚青,黄美玲, 等.计算机辅助3D打印技术用于保留乳房手术一期乳房重建[J].中华乳腺病杂志（电子版）,2018,12(1):12-16. DOI:10.3877/cma.j.issn.1674-0807.2018.01.003

[24] 余广海,吴越,刘斌, 等.大尺度非局部均值伪影抑制技术对提高低剂量腹部CT图像质量的价值研究[J].安徽医科大学学报,2017,52(6):887-891. DOI:10.19405/j.cnki.issn1000-1492.2017.06.024.

[25] 付丽娟,姚宇,付忠良.中值滤波与各向异性扩散相结合的医学图像滤波方法[J].计算机应用,2014,34(1):145-148. DOI:10.11772/j.issn.1001-9081.2014.01.0145.

[26] 刘恋, 周宏, 王晓晶. 基于复四元数小波的图像降噪方法:, 2020.

[27] 张聚,王陈,程芸.小波与双边滤波的医学超声图像去噪[J].中国图象图形学报,2014,19(1):126-132. DOI:10.11834/jig.20140116.

[28] 侯东奥.MRI脑部肿瘤分割及其三维重建[D].江西:南昌航空大学,2018.

[29] 解立志,周明全,田沄,武仲科,王醒策.基于区域增长与局部自适应C-V模型的脑血管分割[J].软件学报,2013,24(08):1927-1936.

[30] 王昌,黄煜峰,王兴家,冯焕清,李传富.最优阈值生长和形态学结合的肺气道树分割方法[J].北京生物医学工程,2010,29(03):241-244+260.

[31] 王雷,高欣,张桂芝.形态膨胀的3D区域生长气管分割算法[J].生物医学工程学杂志,2013,30(04):679-683+691.

[32] 雷继海. 一种分类迭代保边缘图像去噪算法[J]. 自动化与仪器仪表, 2020, No.245(03):35-38.

[33] 张显全, 阳建中, 付念,等. 一种保边缘的图像去噪方法[J]. 计算机工程, 2012, 038(002):224-225.

[34] 李道纪, 张保明, 郭海涛,等. 遥感图像多通道联合辅助降噪技术[J]. 测绘科学与工程, 2018, 038(004):48-53.

[35] 庄天戈. 计算机在生物医学中的应用[M]. 科学出版社, 2000.

[36] 孟祥鹿, 幸子健, 卢山. 基于深度学习的肺结节分类分割算法及其在不同CT重建算法下的效能评估[J]. 中华医学杂志, 2021, 101(07):476-480.

[37] 王光飞, 刘杰, 刘文. 对DICOM 3.0标准下软件工程的研究[J]. 中国医学装备, 2005(07):7-11.

[38] 吴思桦. 基于区块链的医疗隐私保护应用的研究[D].重庆:重庆邮电大学,2019.

[39] 张雨, 黄梦醒, 倪泽浩,等. 一种基于DenseNet的图像噪声识别方法和装置:, 2020.

[40] 陈啸, 边大红, 崔彦宏,等. 基于图像的玉米植株叶倾角概率密度分布函数提取[J]. 国土资源遥感, 2020, v.32;No.125(01):78-83.

[41] 赵本岳. 基于中波广播发射的信道噪声影响分析[J]. 通讯世界, 2019(3):214-215.

[42] 白键, 冯象初. 去除乘性噪声的积分微分方程模型[J]. 西安电子科技大学学报, 2013.

[43] 曹凌. 数字图像处理中噪声源对量化精度影响的分析[J]. 系统工程与电子技术, 1996, 000(011):44.

[44] 蒋辉. 双边滤波理论及其在遥感图像处理中的应用研究[D].成都:西南交通大学, 2014.

[45] 余博, 郭雷, 钱晓亮, et al. 一种新的自适应双边滤波算法[J]. 应用科学学报, 2012(5):517-523.

[46] Xiaoliang Qian,Lei Guo,Bo Yu.Scale-Based Adaptive Smoothing[C].//2010 2nd International Conference on Future Computer and Communication. v.2.2010:40-44.

[47] Ke Chen.Adaptive smoothing via contextual and local discontinuities[J].IEEE Transactions on Pattern Analysis and Machine Intelligence,2005,27(10):1552-1567.

[48] Zhang M. Bilateral filter in image processing[J]. 2009.

[49] Shen-Chuan Tai,Shih-Ming Yang.A Fast Method For Image Noise Estimation Using Laplacian Operator and Adaptive Edge Detection[C].//3rd International Symposium on Communications, Control, and Signal Processing (ISCCSP 2008), vol.3.2008:1077-1081.

[50] 周宇. 基于水平集的图像分割方法研究及其应用[D].西安:西安科技大学 2015.

[51] Vyavahare A J . Canny based DRLSE Algorithm for Segmentation[J]. International Journal of Computer Applications, 2014, 102(7):1-5.

[52] 张帆, 张新红. 基于位错理论的距离正则化水平集图像分割算法[J]. 自动化学报, 2018, 044(005):943-952.

[53] 冉冬梅, 严加勇, 崔崤峣,等. 基于改进DRLSE模型的甲状腺3D超声图像自动分割[J]. 软件, 2019, 040(004):61-66.

[54] 刘国奇, 李晨静. 基于贝塞尔滤波的水平集正则化图像分割方法[J]. 计算机科学, 2018.

[55] YU Jian-feng, LYU Yi-bin, FANG Jin-li,等. 基于Canny算子的C-V水平集模型[J]. 计算机技术与发展, 2019, 029(007):145-149.

[56] Canny J . A Computational Approach To Edge Detection[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 1986, PAMI-8(6):679-698.

[57] 林卉, 赵长胜, 舒宁. 基于Canny算子的边缘检测及评价[J]. 黑龙江工程学院学报, 2003, 17(2):3-6.

[58] 沈凌云, 朱明. 改进的距离正则化水平集演化方法[J]. 北京工业大学学报, 2015(05):40-45.

[59] 段锁林, 殷聪聪, 李大伟. 改进的自适应Canny边缘检测算法[J]. 计算机工程与设计, 2018, 039(006):1645-1652.

[60] Shattuck D W , Sandor-Leahy S R , Schaper K A , et al. Magnetic Resonance Image Tissue Classification Using a Partial Volume Model[J]. neuroimage, 2001, 13(5):856-876.

[61] 徐德明, 万长林. 可抗噪声的Canny改进边缘检测算法 [J]. 计算机系统应用, 2017(1):201-205,共5页.

[62] 杜万萍, 谭利华, 翟宁,等. 颅内血肿体积个人计算机测量方法[J]. 中南大学学报:医学版, 2011.

[63] 李海岩, 阮世捷, 彭翔,等. 一种基于CT图像的颅骨厚度计算机辅助测量方法[J]. 中华创伤杂志, 2007, 23(009):713-714.

[64] 郭俊. 基于FDM3D打印机的温度控制系统设计与切片算法研究[D].成都:西南石油大学, 2017.