-
公开(公告)号：KR1020170028829A
公开(公告)日：2017-03-14
申请号：KR1020160087383
申请日：2016-07-11
Applicant: 삼성전자주식회사
Inventor： 미구킨아르헴세르기비치 , 코로브첸코드미트리알렉산드로비치 , 다닐레비치알렉세이브로니슬라보비치 , 바르폴로메에브나안나안드레에브나 , 리차고브미하일니콜라예비치 , 시로텐코미하일유리에비치 , 최상천
Abstract: 자기공명영상복원을위한방법및 자기공명영상처리장치가제공된다. 자기공명영상처리장치는, 프로세서및 프로세서에연결된메모리를포함하고, 프로세서는, 대상체에대한불완전하게샘플링된멀티코일 k-공간을획득하고, 미리획득된제1 사전및 제1 사전을이용하여획득된제2 사전을이용하여, 멀티코일 k-공간을복원하여대상체의복원영상을획득할수 있다.
-
公开(公告)号：KR101670986B1
公开(公告)日：2016-10-31
申请号：KR1020140031821
申请日：2014-03-18
Applicant: 삼성전자주식회사
Inventor： 굴라카프라빈 , 다니레비치알렉세이브로니스라보비치 , 코로브첸코드미트리알렉산드로비치 , 시로텐코마하일유리에비치 , 우스투자닌드미트리블라디미로비치 , 샤리아메랍아칠에비치 , 리차코브마하일니코라에비치 , 테르노보이세르게이콘스탄티노비치
CPC classification number: G01R33/543 , A61B5/0037 , A61B5/055 , G06K9/20 , G06T7/0012 , G06T2207/10088 , G06T2207/30016
Abstract: 본발명은의료기기분야에관한발명으로서, 특히두뇌의 3차원영상의뷰의자동플래닝에대한것이다. 두뇌의 3차원영상의뷰의자동플래닝방법은, 상기 3차원영상의선택된단면에서계산된기준선들의세트에기초하여두뇌의정중시상(mid-sagittal) 평면의형성, 즉 3차원스카우트-영상을획득하는단계; 상기정중시상면에서검출된랜드마크들을기초로세로기준선을형성하는단계; 상기정중시상면및 세로기준선을기초로방향성을갖는스캔을플래닝하는단계― 상기스캔의플래닝에서, 상기정중시상면의형성을위한기준선들은, 상기 3차원영상의분석에의해선택된단면들의대응하는두 세트의형태로자동으로미리정의된 3차원영상의축상및 관상단면에따라계산되고, 상기영상의픽셀의분석에의해계산된수치기준을이용하여실행된, 상기단면에서뇌의세로갈라짐흔적의검출에기초하여계산되고, 여분의축상및 관상라인들의세트에서, 상기단면에서뇌의세로갈라짐흔적의검출에기초하여계산된기준선들이수집됨―; 대응세트의모든라인의방향의일반화된벡터의방향과양립불가능한라인들의지정된세트로부터예외가되는라인들의세트를여과하는단계; 및정중시상면에서재형성된영상및 이영상에서검출된랜드마크들에기초하여세로기준선을형성하는단계를포함한다.
-
公开(公告)号：KR101645292B1
公开(公告)日：2016-08-03
申请号：KR1020140031156
申请日：2014-03-17
Applicant: 삼성전자주식회사
Inventor： 굴라카프라빈 , 다니레비치알렉세이브로니스라보비치 , 코로브첸코드미트리알렉산드로비치 , 리차코브마하일니코라에비치 , 시로텐코마하일유리에비치
CPC classification number: G06T7/0012 , A61B6/032 , A61B6/5223 , G06F19/00 , G06T7/75 , G06T7/77 , G06T19/20 , G06T2207/10072 , G06T2207/20101 , G06T2207/20124 , G06T2210/41 , G06T2219/008 , G06T2219/028 , G09B23/28 , G09B23/30
Abstract: 본발명은의료기기에관계되며, 특히의료영상들의분석기술에관한것이다. 본발명은 3차원의료영상들의뷰의자동플래닝방법으로서, 해부학적지점들의위치들의통계모델의추정, 해부학적지점들의검출트레이닝을수행, 관심영역의 3차원영상을획득, 해부학적지점들의후보들의집합의검출을수행, 해부학적지점들의좌표들의집합의검출을수행, 후보들중 해부학적지점들의최적구성의탐색을수행, 형성된해부학적지점들의구성에기초하여시야면의형성을수행하고, 상기통계모델의파라미터들의추정은 3차원영상들의마킹된(annotated) 집합을사용하여수행되고, 상기해부학적지점들의검출트레이닝은손실함수의일부인통계모델의에너지를사용하여수행되고, 상기해부학적지점들의후보들의집합의검출은위 알고리즘을사용하여트레이닝된해부학적지점들의검출부에의해수행되고, 상기해부학적지점들의최적구성의탐색은통계모델의특정에너지를최소화하는구성의그리디반복탐색수단에의해수행된다.
Abstract translation: 提供医疗设备和分析医学图像的技术。 用于在三维（3D）医学图像中自动规划视图的方法包括：估计指示解剖点的位置的统计模型，所述统计模型具有通过最小化损失函数的能量计算的参数; 训练解剖点检测器以通过使用估计的统计模型来检测多个解剖点; 获取具有感兴趣区域的3D图像; 检测所述3D图像中的解剖点的一组候选者; 搜索所述候选集合以获得对应于所述多个解剖点的最佳配置; 并且基于通过搜索找到的最佳配置来形成视平面。
-
Patent Agency Ranking
公开(公告)号：KR1020140114303A
公开(公告)日：2014-09-26
申请号：KR1020140031156
申请日：2014-03-17
Applicant: 삼성전자주식회사
Inventor： 굴라카프라빈 , 다니레비치알렉세이브로니스라보비치 , 코로브첸코드미트리알렉산드로비치 , 리차코브마하일니코라에비치 , 시로텐코마하일유리에비치
CPC classification number: G06T7/0012 , A61B6/032 , A61B6/5223 , G06F19/00 , G06T7/75 , G06T7/77 , G06T19/20 , G06T2207/10072 , G06T2207/20101 , G06T2207/20124 , G06T2210/41 , G06T2219/008 , G06T2219/028 , G09B23/28 , G09B23/30 , G06T17/10
Abstract: The present invention relates to a medical device, and more particularly to a technology of analyzing medical images. The present invention provides a method of automatically planning views in a 3-D medical image, which comprises the steps of: estimating statistics models of locations of anatomic points; performing a detection training of the anatomic points; acquiring a 3-D image of an interest area; detecting a set of candidates of the anatomic points; detecting a set of coordinates of the anatomic points; searching for optimum configurations of the anatomic points among the candidates; and forming a field of view surface based on the configurations of the formed anatomic points, wherein the parameters of the statistics models are estimated by using an annotated set of 3-D images, a detection training of the anatomic points is performed by using energy of the statistical models, which are parts of a loss function, the set of the candidates of the anatomic points are detected by detection units of the trained anatomic points, and the optimum configurations of the anatomic points are searched for by a grid repetition searching unit for the configuration of minimizing specific energy of the statistical model.
Abstract translation: 医疗装置技术领域本发明涉及医疗装置，更具体地，涉及医学图像分析技术。 本发明提供了一种自动规划三维医学图像中的视图的方法，包括以下步骤：估计解剖点位置的统计模型; 对解剖点进行检测训练; 获取兴趣区域的三维图像; 检测一组解剖点的候选者; 检测解剖点的一组坐标; 寻找候选人之间解剖点的最佳配置; 并且基于所形成的解剖点的构造形成视野表面，其中通过使用注释的3-D图像集来估计统计模型的参数，通过使用能量来进行解剖点的检测训练 作为损伤函数的一部分的统计模型，通过训练解剖点的检测单元检测解剖点候选的集合，并且通过网格重复搜索单元搜索解剖点的最佳配置， 最小化统计模型的比能量的配置。
-
公开(公告)号：KR101599219B1
公开(公告)日：2016-03-03
申请号：KR1020140031858
申请日：2014-03-18
Applicant: 삼성전자주식회사
Inventor： 굴라카프라빈 , 다니레비치알렉세이브로니스라보비치 , 코로브첸코드미트리알렉산드로비치 , 리차코브마하일니코라에비치 , 시로텐코마하일유리에비치
CPC classification number: G06T7/0012 , G06T7/33 , G06T2200/04 , G06T2207/10072 , G06T2207/10081 , G06T2207/10088 , G06T2207/10104 , G06T2207/20024 , G06T2207/20076 , G06T2207/20081 , G06T2207/30204
Abstract: 본발명은의료기기에관한것이며, 특히의료영상들상의해부학적지점(랜드마크)들의자동등록에관한방법및 디바이스에관한것이다. 이는 3차원의료영상들의랜드마크들의자동등록방법에관한것이며, 3차원의료영상의획득, 탐색지점들의집합의결정, 지점들의특징의추출, 필요한랜드마크들상의후보들의집합의형성, 후보들의필터링, 필요한랜드마크들의최종위치의출력과같은동작의수행을제공한다. 여기에서탐색지점들의설정의결정은통계지도의사용을갖는 3차원영상내에서탐색지점들의그리드의설정에의해수행되고, 통계지도는 3차원영상내부의신체의조사된부분에대응하는경계박스에부착되고, 지점들의특징의추출은이러한각 지점에대한주변컨텍스트의선택및 특징추출다층알고리즘의적용에의해수행되고, 상기후보들의집합의형성은특징의추출알고리즘의대응하는출력값을사용하여필요한랜드마크들의각 유형및 각탐색지점에대한품질의측정의계산에의해수행되고, 상기후보들의필터링은품질측정값에대한미리계산된임계값들및 통계지도의수단에의해수행된다. 여기에서최종위치의출력은필터링후에남아있는모든후보들을필요한랜드마크들의종류별제한내에서정렬하고, 품질측정치의가장큰 값을갖는후보들또는품질측정치의가장큰 값을갖는필요한수의후보들을출력하는것에의해서수행된다. 청구된방법의구현에기초하는장치및 데이터캐리어도또한본발명의일부이다.
Abstract translation: 本发明涉及一种医疗装置，特别涉及用于在医学图像上自动登记解剖点（地标）的方法和装置，其中使用在3D医学图像中自动登记地标的方法，提供获得3D 确定一组搜索点，提取搜索点的特征，根据要求的地标形成一组候选人，过滤候选人，以及输出所要求的地标的最终位置。 搜索点的确定通过使用统计图集在3D图像上设置搜索点的网格来执行，统计图谱被附加到与3D图像内的身体的被调查部分相对应的边界框，特征提取 通过选择每个点周围的上下文并对其应用多层特征提取算法来执行搜索点的执行，通过使用相应的输出来计算每个搜索点和每种类型的地标的质量测量来执行一组候选的形成 的特征提取算法，并且通过使用统计图谱和预先计算的阈值来执行候选的滤波。 这里，最终位置是通过根据质量测量对每种类型的所需地标的限制内的所有候选者进行排序，并且通过输出质量测量值最大值的候选或输出最大值的候选数 质量措施。 基于所要求保护的方法的实施例的装置和数据载体也是本发明的一部分。
-
公开(公告)号：KR1020150125543A
公开(公告)日：2015-11-09
申请号：KR1020140195373
申请日：2014-12-31
Applicant: 삼성전자주식회사
Inventor： 미구킨아르헴세르기비치 , 코로브첸코드미트리알렉산드로비치 , 시로텐코미하일유리에비치 , 가브릴유크키릴알투로비치 , 굴라카프라빈 , 최상천 , 리차고브미하일니콜라예비치 , 최양림
CPC classification number: G06T7/0012 , A61B5/055 , A61B5/7257 , A61B2576/00 , G01R33/5611 , G06T11/003 , G06T2207/10088 , G06T2207/20081 , A61B5/0033 , A61B5/004
Abstract: 자기공명영상장치는 k-공간상의언더샘플링된스펙트럼을획득하는획득부; 및상기언더샘플링된스펙트럼을통하여목적영상을생성하는복구부를포함하고, 상기복구부는, 상기 k-공간상의비-샘플링된위치에대응하는데이터에대해서분할브레그먼기법또는근사스파스부호화기법을통하여초기복구를수행하는제1 서브복구부; 상기 k-공간상의상기초기복구된스펙트럼을다수의개별스펙트럼들로다중대역분해하고, 상기분해된대역에대응하는영상들에대해서측정된주파수를스파스(sparse) 근사화하고복구하는것을교번하여사전-기반(dictionary learing) 복구를하는제2 서브복구부; 및상기각 주파수대역의복구된영상을병합하여, 목적영상을생성하는영상생성부를포함할수 있다.
Abstract translation: 磁共振成像装置包括：获取单元，用于在k空间上获取欠采样谱; 以及恢复单元，用于通过欠采样频谱产生目标图像，其中恢复单元可以包括：第一子恢复单元，用于通过分割Bregman技术对与k空间上的非采样位置相对应的数据执行初始恢复 或近似稀疏编码技术; 第二子恢复单元，其在多个频带中将k空间上的初始恢复的频谱分解成多个单独的频谱，并且通过交替稀疏近似和恢复针对对应于图像的图像测量的频率来执行字典学习恢复 分解带; 以及图像生成单元，用于合并各个频带的恢复图像，以便生成目标图像。
-
公开(公告)号：KR1020140114308A
公开(公告)日：2014-09-26
申请号：KR1020140031858
申请日：2014-03-18
Applicant: 삼성전자주식회사
Inventor： 굴라카프라빈 , 다니레비치알렉세이브로니스라보비치 , 코로브첸코드미트리알렉산드로비치 , 리차코브마하일니코라에비치 , 시로텐코마하일유리에비치
CPC classification number: G06T7/0012 , G06T7/33 , G06T2200/04 , G06T2207/10072 , G06T2207/10081 , G06T2207/10088 , G06T2207/10104 , G06T2207/20024 , G06T2207/20076 , G06T2207/20081 , G06T2207/30204
Abstract: The present invention relates to a medical device, and more particularly to a method and a device for automatically registering anatomic points (landmarks) on medical images. The present invention relates to a method of automatically registering landmarks in 3D medical images, and provides operations of acquiring a 3D medical image, determining a set of search points, extracting features of the points, forming a set of candidates on the necessary landmarks, filtering candidates, and outputting final locations of the necessary landmarks. Here, setting of search points is determined by setting grids of search points in a 3D image having use of a statistical map, a statistical map is attached to a border box corresponding to an irradiated portion of the body in the interior of the 3D image, features of the points are extracted by selecting a peripheral context and applying a feature extraction multilayer algorithm, the set of the candidates is formed by calculating types of necessary landmarks and the measurement of qualities of the search points using an output value corresponding to an extraction algorithm of the feature, and the candidates are filtered by premeasured threshold values for the quality measurement values and a means of a statistical map. Here, the final location is output by arranging all candidates left after the filtering in limits for types of necessary landmarks, and outputting candidates having the largest value of quality measurement values and a necessary number of candidates having the largest value of quality measurement values. An apparatus and a data carrier based on realization of the claimed method also are parts of the present invention.
Abstract translation: 医疗装置技术领域本发明涉及一种医疗装置，更具体地涉及一种用于自动登记医学图像上的解剖点（地标）的方法和装置。 本发明涉及一种在3D医学图像中自动登记地标的方法，并提供获取3D医学图像的操作，确定一组搜索点，提取点的特征，在必要的界标上形成一组候选，过滤 候选人，并输出必要地标的最终位置。 这里，通过在使用统计图的3D图像中设置搜索点的网格来确定搜索点的设置，统计图被附加到与3D图像的内部的身体的照射部分相对应的边框上， 通过选择外围环境和应用特征提取多层算法来提取点的特征，通过使用与提取算法相对应的输出值来计算所需的地标的类型和搜索点的质量的测量来形成候选的集合 的特征，并且通过用于质量测量值的预测阈值和统计图的手段来过滤候选物。 这里，最终的位置是通过排列所有必需地标类型的限制之后的滤波所剩余的候选者，输出质量测量值的最大值的候选和具有最高的质量测量值的候选的候补。 基于所要求保护的方法的实现的装置和数据载体也是本发明的一部分。
-
公开(公告)号：KR1020140114307A
公开(公告)日：2014-09-26
申请号：KR1020140031821
申请日：2014-03-18
Applicant: 삼성전자주식회사
Inventor： 굴라카프라빈 , 다니레비치알렉세이브로니스라보비치 , 코로브첸코드미트리알렉산드로비치 , 시로텐코마하일유리에비치 , 우스투자닌드미트리블라디미로비치 , 샤리아메랍아칠에비치 , 리차코브마하일니코라에비치 , 테르노보이세르게이콘스탄티노비치
CPC classification number: G01R33/543 , A61B5/0037 , A61B5/055 , G06K9/20 , G06T7/0012 , G06T2207/10088 , G06T2207/30016
Abstract: The present invention relates to a medical device, and more particularly to automatic planning of a view in a 3D image of a brain. The method of automatically planning a view in a 3D image of a brain comprises the steps of: forming a mid-sagittal plane of a brain based on a set of reference lines calculated from a selected section of the 3D image, that is, acquiring a 3D scout image; forming a longitudinal reference line based on landmarks detected from the mid-sagittal plane; planning a scan having a directionality based on the mid-sagittal plane and the longitudinal reference line-in the planning of the scan, reference lines for forming the mid-sagittal plane are calculated according to an axial shape and a coronal section of the 3D image automatically predefined in the form of corresponding two sets of the sections selected through analysis of the 3D image, calculated based on detection of a longitudinally cracked trace of the brain in the section, executed using a numeral reference calculated through analysis of a pixel of the image, and calculated based on detection of the longitudinally cracked trace of the brain in the section, in a set of marginal axial and coronal lines-; filtering a set of lines excluded from a designated set of lines which are incompatible with a direction of a generalized vector of directions of all lines of the corresponding set; and forming a longitudinal reference line based on image reformed on the mid-sagittal plane the landmarks detected from the image.
Abstract translation: 医疗装置技术领域本发明涉及一种医疗装置，更具体地涉及脑的3D图像的视图的自动规划。 自动规划大脑3D图像中的视图的方法包括以下步骤：基于从3D图像的所选部分计算的参考线集合，形成脑的中矢状面，即，获取 3D侦察图像; 基于从中矢状面中检测到的界标形成纵向参考线; 根据中间矢状面和纵向参考线计划具有方向性的扫描 - 在扫描的规划中，根据3D图像的轴向形状和冠状面计算用于形成中矢状面的参考线 通过分析3D图像而选择的相应的两组部分的形式自动预定义，所述三维图像通过使用通过分析图像的像素而计算出的数字参考，基于检测到该部分中的脑的纵向裂纹痕迹而计算 ，并且基于在一组边缘轴向和冠状线中的部分中的脑的纵向裂纹痕迹的检测来计算; 从与所述对应集合的所有行的广义向量的向量的方向不兼容的指定的一组行中排除的一组行的过滤; 并且基于在中矢状面上重新形成从图像中检测到的地标的图像形成纵向参考线。
-
-
-
-
-
-
-
Search Results
8
records
(took 0.04 seconds)
