運用形態學技術於細胞個數計算之行動裝置應用

科技不斷的進步，人們對健康日益重視，對生物檢測的需求也日益增加，舉凡血液細胞檢測、癌細胞檢測…等。傳統的人工檢測，由實驗人員用肉眼計數觀測，檢測結果易受實驗人員的生理狀況或心理狀況影響，而造成結果有誤差。細胞計數自動化，是將人工計數的部分由電腦來計算，降低人力成本支出，也有較穩定的結果。本論文實作一個簡易的可攜式細胞計數系統，結合了樹莓派(Raspberry Pi)、Pi 攝影設備、μ Handy行動顯微鏡與影像處理演算法，主要是運用高斯模糊、OTSU影像二值化及形態學運算等影像處理演算法來處理影像，以達到計數細胞的目的。本論文所開發的細胞計數應用程式，使用者可以依照細胞圖片的特性做相關的演算法調整，以降低誤差率。實驗的結果顯示計數正確率可達95%以上。

關鍵字

樹莓派；細胞計數；形態學

並列摘要

As technology continues to advance, the need of biological testing such as detection of blood and cancer cells increase as people pay more attention to their health concern. Conventional manual biological testing, cells are counted one by one with eye observation of operators, however, the physical and mental conditions of the operators are likely to affect the observation test to be erroneous results. An automated cell counting by computer is a solution to replace manual counting which can reduce the labor cost and have less error-prone results. In this thesis, an automated cell counting system, that combines a Raspberry Pi, Pi camera, μ Handy action microscopy and image processing algorithms, including the processes of image Gaussian blur processing algorithm, OTSU binary image and morphological operations, is implemented to perform the cell counting in a mobile device. With the implemented automatic cell counting system, users can adjust the algorithm in accordance with the properties of cell image to lower the counting error rate. Experimental results show that our system can achieve accuracy higher than 95%.