颜光前, 赵柳, 吴俊, 陈悦, 陈林, 裘之瑛. 基于ABUS图像的轻量型切口疝补片计算机辅助检测与评估算法[J]. 云南大学学报(自然科学版), 2017, 39(5): 768-779. doi: 10.7540/j.ynu.20160741
引用本文: 颜光前, 赵柳, 吴俊, 陈悦, 陈林, 裘之瑛. 基于ABUS图像的轻量型切口疝补片计算机辅助检测与评估算法[J]. 云南大学学报(自然科学版), 2017, 39(5): 768-779. doi: 10.7540/j.ynu.20160741
YAN Guang-qian, ZHAO Liu, WU Jun, CHEN Yue, CHEN Lin, QIU Zhi-ying. Computer-aided detection and evaluation algorithm of lightweight incisional hernia mesh based on ABUS images[J]. Journal of Yunnan University: Natural Sciences Edition, 2017, 39(5): 768-779. DOI: 10.7540/j.ynu.20160741
Citation: YAN Guang-qian, ZHAO Liu, WU Jun, CHEN Yue, CHEN Lin, QIU Zhi-ying. Computer-aided detection and evaluation algorithm of lightweight incisional hernia mesh based on ABUS images[J]. Journal of Yunnan University: Natural Sciences Edition, 2017, 39(5): 768-779. DOI: 10.7540/j.ynu.20160741

基于ABUS图像的轻量型切口疝补片计算机辅助检测与评估算法

Computer-aided detection and evaluation algorithm of lightweight incisional hernia mesh based on ABUS images

  • 摘要: 由于人工检阅自动化三维乳腺超声(ABUS)图像极其耗时,而且极易出现对微弱异常区域的漏诊.为了提高ABUS图像检阅效率并减少漏诊,提出一套基于ABUS图像的轻量型切口疝补片计算机辅助检测与评估算法.算法首先使用纹理特征萃取算法自动量化提取三维感兴趣容积中的待分类区域的相关纹理特征参数,以便用于对补片和筋膜的区分;然后,针对二维纹理参数对切口疝补片术后卷曲、收缩等空间变换较为敏感的问题,引入三维纹理参数和三维位置参数来提高轻量型补片分类识别算法的鲁棒性;最后,使用类间距算法和顺序前进搜索法来进行特征选择,并使用支持向量机进行分类识别.算法可有效降低人工阅片的工作强度,辅助医生识别ABUS扫描区域内有无轻量型补片,并对补片相关诊断项目做出辅助评估.

     

    Abstract: Due to that reading Automated 3D Breast Ultrasound (ABUS) images is time consuming,and subtle abnormalities may be missed.To make reading more efficient and to reduce reading errors,a computer-aided detection and evaluation algorithm is proposed to automatically find implanted lightweight meshes in ABUS images.First,a textural feature extraction algorithm is presented to automatically find candidate objects in the volume of interest and compute textural features on multiplanar images for classification of the mesh and fascia.Second,the 2D texture is more sensitive to the mesh shrinkage.Therefore,3D texture and 3D position parameters are introduced to enhance the robustness of the implanted meshes recognition method.Finally,the distance between class algorithm and a sequential forward selection algorithm are used for the feature selection.And the support vector machine is used to classify and distinguish.The algorithm can effectively reduce the intensity of reading,and assists the doctor to identify the presence of lightweight patch in the ABUS scanning area.And it can assist in the evaluation of patch-related diagnostic programs.

     

/

返回文章
返回