Workshop - IMIC
MICCAI Daily: Thursday
During Monday’s workshop about Interactive Medical Image Computing
gave a talk about his paper "Interactive Tracking of
Cells in Microscopy Image Sequences".
This work has the goal of providing a reliable tool for live cell tracking in
microscopy videos with the supervision of a domain expert.
The proposed approach relies on level set segmentation algorithms to keep
track of the boundary of each cell in every frame of the video; this is
necessary in order to extract useful metrics for medical analysis (e.g. the
measure of the surface of a cell or its movement over the substrate).
However automated algorithms make mistakes, therefore an expert is
asked to oversee the tracking process and he is given the tools to stop the
tracking and manually correct the mistaken cell boundaries.
Furthermore the research group also proved that a K-Nearest Neighbors
classifier can be trained in order to predict when the expert intervention is
For future work, the team plans on involving inexpensive crowd workers
with the purpose of fixing the algorithm mistakes without the expert
Interactive Tracking of Cells in
Microscopy Image Sequences