CIA Lab Technologies - Real-time Demonstration
A real-time demonstration of our lung lesion segmentation
As demonstrated by the real-time movie below, an operator manually initializes the algorithm by specifying an elliptical region-of-interest enclosing the lesion on one slice, and the algorithm will then automatically segment the lesion in 3D. The red contours superimposed on the lesion are generated by the algorithm.
Highlights of our lesion and organ segmentation algorithms:
An example of speed: A liver case consisting of 76 MR images/slices took our liver organ algorithm 23.7 seconds to complete for volumetric segmentation. There was no need for manual correction.
An example of accuracy: we received ~87% (Dice) mean overlap ratio for our lung lesion segmentation algorithm when applying it to the publicly accessible lung datasets of same-day repeat CT and LIDC.