Deceived bilateral filter for improving the automatic cell segmentation and tracking in the NF-kB pathway without nuclear staining

Abstract

In this work we implement an automatic videobased online cell tracking system for TNFa treated cells in the transcription factor NF-kB pathway, with the aim to provide p65 measurement approximation with time and individual cell resolution. The video used to test our proposed tracking system consists on cells transfected with fluorescent p65 protein. Such protein switches its location within the cell, among the nucleus and cytoplasm over time, causing important changes on its apparent morphology. This represents a major challenge for an automatic cell tracking system. Additionally, images in the tested microscopy footage present noise and contrast variations. For noise reduction, we modified the deceived bilateral filter (DBF), a denoising and edge sharpening non linear filter. The proposed modification of the DBF consists in an adaptive border enhancement. We tested the cell tracking algorithm with different preprocessing configurations achieving a maximum tracking accuracy improvement of 28%, with the use of the modified DBF