Instead of trying to select the optimum scale for each edge, and then segmenting the image on the selected edges, this approach collects edge information at multiple scales and then does a simultaneous segmentation over all the edges.
Not surprisingly, the flux-tensor segmentation tends to over-segment cells, i.e., split each cell into several segments, while under-segmentations are observed less frequently.
These values were selected empirically to achieve the best segmentation results over all patients.
The mean absolute segmentation error over all A-Scans lies below 10 µm on 95.1% of the images.
This finally allowed us to compare the results of the topological alignment with the annotated segmentation by counting over-, under-, and mis-partitionings.
Instead of segmenting images via conventional techniques, inspired by the idea of the ensemble clustering technique that combines a set of weak clusterers to obtain a strong clusterer, we propose to achieve a consensus segmentation by fusing evidence accumulated from multiple weak segmentations (or over-segmentations).
We then collected and evaluated each recorded mask against a manually segmented ground truth, thus allowing us to gauge segmentation accuracy over time.
A new animation clip (see Supplementary material) explains all steps, from segmentation, over surface-reconstruction to landmark-positioning, within the software package Amira, hence helping to apply this method to comparable material and scientific fields.
Note that L m,n is the sum of all the within-segment variances and g(m,n) increases when we add more segments, which penalizes over-segmentation.
The method performs segmentation accuracy over 90%.
