Logo image
Quantitative Analysis of Interstitial Cells of Cajal in Circular Muscle of Human Colon
Thesis

Quantitative Analysis of Interstitial Cells of Cajal in Circular Muscle of Human Colon

Kuo, Teng-Fang
Masters, 國立清華大學, 電機工程學系
2011

Abstract

定量分析 Cajal間質細胞 三維模型 神經追蹤 Quantification Interstitial Cells of Cajal 3D model Neuron tracing
Interstitial Cells of Cajal(ICC) is the pacemaker of motility in the gastrointestinal tract. It is found in recent years that its amount and processes are good indicators to characterize some motility disorders such as constipation and diabetes. In fact, quantitative analysis of ICC is an important topic nowadays. Traditionally, doctors quantify the amount of ICC slice by slice by human eyes, which is accurate but tedious. After optical clearing technique proposed [1], now we are able to acquire images from deeper depth in the specimens. However, additional amount of images acquired by the new technique has made the traditional method inapplicable. Therefore, a more efficient method is required. Few attempts have been made at the amount counting of ICC. One proposed method counts ICC in projection images in x-y, x-z, and y-z planes and back projecting them to 3-D image. However, using this method encounters the difficulties in distinguishing ICCs from each other when they are all projected to a single image in our case. We present a method called “centroid classification” which adopts the property that centroids of ICCs over slices would be located within a reasonable region to quantify ICCs. Another research topic for ICCs is the length counting of ICC. There are two main directions in this study: 1) 3-D thinning and 2) finding the medial line. We propose a new method to solve the inefficiency of the previous proposed methods based on finding the overlap ratio of the same ICC process in adjacent images, and using this information to reconstruct the 3-D structure. In our experiment, the two methods we proposed are efficient with acceptable accuracies which would be helpful in clinical diagnosing.

Metrics

1 Record Views

Details

Logo image