New work from our lab pushes forward the performance limits of unsupervised methods in the difficult task of segmenting cells and their nuclei in tissue samples in the context of immunofluorescence imaging and its highly multiplexed counterpart. As part of UNSEG, we have a developed a new perturbed watershed method that stabilizes and improves classical watershed performance for segmenting clustered nuclei.
B. Kochetov, P. Bell, P. S. Garcia, A. S. Shalaby, R. Raphael, B. Raymond, B. J. Leibowitz, K. Schoedel, R. M. Brand, R. E. Brand, J. Yu, L. Zhang, B. Diergaarde, R. E. Schoen, A. Singhi, S. Uttam