Cell Density Estimation from a Still Image for In-Situ Microscopy

Cell counting based on local intensity maxima grouping for in-situ microscopy

“…Various algorithms for cell counting can be found in professional literature, and they can be roughly classified in two groups: those which assume that the cells in the scene do not build clusters [3,5], and those which assume that the cells in the scene do build clusters [2,7,8,4,9,10,6]. For those algorithms which assume that the cells do not build clusters, each cell region in the image is segmented.…”

“…For on-line estimation of the cell density (cell count per volume unit) inside of a bioreactor, a steam-sterilizable bright field in-situ microscope is being investigated [1,2]. An insitu microscope has a video camera incorporated to its structure, as well as a sampling zone that can be inserted inside the bioreactor.…”

“…Taking into account the previously mentioned characteristics, a new cell counting algorithm will be developed. First, it will segment the cell cluster regions by applying the global thresholding algorithm proposed by [2]. Then, the local intensity maxima inside each cell cluster region will be grouped by taking into account their relative location and intensity values.…”

“…Figure 1 depicts a real intensity image of Baby Hamster Kidney (BHK) cells captured by an in-situ microscope. The cell density at any given time is estimated from an intensity image captured at that time via image processing algorithms [2]. First, the cell clusters regions are found (segmented).…”

“…For those algorithms which assume that the cells do build clusters, each cell cluster region in the image is segmented. Then, the number of cells in each cluster region is estimated by dividing the cell cluster area by the area of a single cell [9,2], or by segmenting either the cells [4,8,6] or the cell nuclei [7,10] inside each cluster. Finally, the total number of cells in the image equals to the sum of the amount of estimated cells in each cluster.…”