AN AUTOMATIC NEURON RECONSTRUCTION SYSTEM
AN AUTOMATIC NEURON RECONSTRUCTION SYSTEM
复制标题
DOI：
10.1016/0165-0270(83)90092-4
复制
发表时间：
1983-01-01
影响因子：
3
通讯作者：
CAPOWSKI, JJ
中科院分区：
文献类型：
--
作者：
CAPOWSKI, JJ
大多数现代神经元追踪系统都是半自动的；也就是说，操作员在显微镜或数据板上追踪神经元分支，并指示计算机何时记录空间样本。然后，使用这样的样本列表来生成神经元的显示并以统计方式总结其结构。跟踪很繁琐，操作员必须对点类型做出主观决定，因此会对结果产生偏差。建立了一个计算机通过电视数字化系统观看图像的全自动系统。计算机跟踪纤维并识别分支点和纤维末端。坐标、点的类型和纤维直径被记录在计算机存储器中。操作员只需要开始跟踪光纤，并在计算机变得混乱时帮助计算机。结果好坏参半。计算机可以非常成功地跟踪纤维，并在一定程度上成功地确定分支和末端，但这一过程比熟练操作员使用精心设计的半自动系统要慢一个数量级。
Most contemporary neuron tracing systems are semi-automatic; that is, an operator follows neuron branches in a microscope or on a data tablet and instructs a computer when to record a spatial sample. A list of such samples is then used to generate a display of the neuron and to summarize statistically its structure. The tracing is tedious, and the operator must make subjective decisions about point types, thus biasing the results. A fully automatic system in which the computer views the image through a television digitizing system has been constructed. The computer follows the fibers and recognizes branchpoints and fiber endings. The coordinates, types of points, and fiber diameters are recorded in computer memory. An operator is required only to start the tracing of a fiber and to help the computer when it becomes confused. The results are mixed. The computer can follow fibers quite successfully and can determine branches and endings with some success, but the procedure is an order of magnitude more slow than that of a skilled operator using a well-designed semi-automatic system.