OBJECTIVE: To develop a novel automated image analysis system to differenti
ate immunohistochemically stained cells from background.
STUDY DESIGN: Cell segmentation was performed by applying global thresholdi
ng algorithms to find an approximate threshold at which cells could be sepa
rated from background followed by a novel refinement algorithm to erode edg
e pixels of the region. To separate overlapping cells, a new decomposition
method was developed that uses both semantic knowledge and high-level relat
ional information. Both the cell segmentation and separation methods were e
valuated on images of stained tissue sections and the manually outlined cel
l areas and numbers compared to the computed.
RESULTS: Macrophage areas computed at the first stage by Otsu's algorithm d
id not differ significantly (P = .07) from those traced manually, while the
areas computed by Kittler's and Kurita's algorithms did not agree (P < .01
). Both Otsu's and Kurita's algorithms performed well when combined with ed
ge pixel erosion. Kittler's algorithm proved unsuccessful even with edge er
osion. Comparison of the computed and manually determined cell numbers show
ed a significant con elation, and regression analysis resulted in the unity
curve.
CONCLUSION: A combination of global thresholding and a novel edge erosion t
echnique allowed identification of immunohistochemically stained macrophage
s; the computed cell areas agreed with the manual results.