Font Size: a A A

Research Of Cell Detecting, Counting And 3-D Reconstruction From Microscopic Images Based On Convergence Index

Posted on:2016-12-06Degree:DoctorType:Dissertation
Country:ChinaCandidate:D SuiFull Text:PDF
GTID:1108330479978640Subject:Computer application technology
Abstract/Summary:PDF Full Text Request
Microscopy images processing are very important in the field of medica l care as well as the life sciences. With the development of microscopic techniques, re-searchers have developed many microscopy platforms for different usage, which generated lots of complex biological image data sets to assist experimental analysis. The increasing of biological micro images with high content and high information bring great challenges to researchers in computer science and biology.The concerned research, cell detection and counting, three dimensional recon-struction of neuronal cells anatomical structure have play important roles in cell based re-searches. There already were many algorithms and software packages which have been developed for the purposed of improving the efficiency, but these methods were only offered as auxiliary, and have many limitations.In this dissertation, the research on cell detection, counting and reconstruction was carried out. Density packed cells’ and insect cells’ detection and counting methods were developed for certain purposes, and then a preliminary resear ch on neuron cells’ detection and anatomy structural reconstruction methods were per-formed. The main contributions of the dissertation are summarized as follows:We proposed a density packed cell detection and counting method based on a sliding band filter. And then the proposed method was applied to the density packed cells in ONL(Out Nuclear Layer) of cat retinal for cells detecting and counting. The results show our proposed method exhibited an excellent performance with its ac-curacy compared with human manual counting and with higher accuracy than tradi-tional methods. Furthermore, the proposed method can aid the pathological diagno-sis and depicted the relationship between the changes in cell number and physio-logical activity. It is worth noting that the proposed cell counting method can clearly benefit for retinal detachment and reattachment visual diagnostics close related to cell loss and addition.We explored a new cell counting method for sf9 cells in bright field for the purposes of solving the cell counting problem in the large scale culture of bacu-lvirus expression system. We first compared the gradient vector distribution in the collected images between bright and dark field, and a transformed sliding band filter was proposed by modifying the calculating method of convergence index to en-hance the center of cells. And then the cell number was generated by the searched cell center. The results exhibit an excellent performance with its high accuracy in lower error rate compared with traditional methods and manual counting. With the superior performance of the experimental results, it is proven that the proposed in-sect cell counting method can clearly improve the efficiency of baculvirus expres-sion system.Neuron cell anatomy structural reconstruction plays a very important role in the field of neurology. In this part, we extended the 2 dimension convergence index and sliding band filter into three dimension, and proposed a sliding volume filter for seeds detection. We first get a series of critical points by sliding volume filter from neuron cell image datasets. And then seed points was selected by the rule of ridges based on the critical points. The results show our proposed methods was better than the traditional ones.In the part of neuron anatomy reconstruction, we modify the traditional open curve snake model by introducing a new external force to aid the curve evolution for neuron skeleton reconstruction. Following that, we introduced a radius estima-tion method based on 2D sliding band filter to extract the cross section contour. Then we finished the anatomy structure reconstruction of neuron cells by radius contour line reconstruction. In the neuron anatomy reconstruction part, we proposed a series of improved methods on seeding, skeleton reconstruction and radius esti-mation, and presented the whole neuron anatomy structure by means of contour line reconstruction. The results show our proposed protocol can clearly benefit for neu-ron anatomy reconstruction on seeding, skeleton reconstruction under ima ge signal reduction and radius estimation.In this dissertation, some exploratory researches of microscopic image analysis on 2-D cell counting and 3-D reconstruction of neuron cell anatomy structural were carried out. All of these results demonstrated that our proposed method can be ap-plied to microscopic images analysis for understanding cells’ number and anatomy information.
Keywords/Search Tags:Microscopic image processing, Cell counting, Neuron reconstruction, Convergence index, Open curve snake model
PDF Full Text Request
Related items