Histological studies of nerve cells are used by researchers to explore many nerve pathologies and treatments. Manual analysis and interpretation of hundreds of nerve fibers typically seen on one image is not possible and therefore, we propose a semi-automated tool based on ridge detection to allow users to segment histological images as they desire. An automatic segmentation is used to first segment the image. Ridges of the cells were enhanced using the first eigenvalue of the hessian matrix. A hysteresis based ridge segmentation is then used to separate the cells from the background. The completed automatic segmentation is presented to the user. It can be further enhanced to meet the individual needs of the user. The user has the ability to (1) remove any cells in the histological image that are not desired cells or (2) add any cells not initially detected. Once the user-defined segmentation is completed, the final segmented image is used to analyze the properties of each individual cell. The size of each cell as well as a measure of wall thickness is calculated and saved for the user. In images of good contrast, where nerve cells are large and well spaced, the automatic cell detection rates are between 90 - 95%. In images with poor contrast with small nerve cells tightly packed together, the cell detection rate is 50-75%.
Mendeley saves you time finding and organizing research
Choose a citation style from the tabs below