Ranking is based on the following 4 rules and their respective weightage:
- Number Measure (Weightage 25%): (NM)
- Deviation Measure (Weightage 25%): (DM)
- Roundedness (Weightage 20%): (RD)
- Report (Weightage 30%): (RT)
The expert annotation of segmented cells has an average segmentation result of 150 cells per High Frequency Fields (image).
Let S be the average of the segmented cells in the dataset, then
The localisation quality of the segmentation method would be evaluated on the basis of the average deviation of the segmented cell centroids from the centroids of the corresponding ground truth cells. This measure is given by:
where n is the number of correctly detected cells and d_k is the distance of the k^th (k ε 1, 2, 3, ……. , n) correctly detected cell centroid from the centroid of the corresponding ground truth cell. The lower the value of this parameter, the better the segmentation performance. The participants (suppose there are X participants) would be ranked on the basis of this parameter and the ranks [1 X] would be linearly mapped onto [25 10] which will be the value of the DM parameter.
Let circularity measure of a segmented blob be the ratio of the Area of the shape to the perimeter square
The value of RD parameter will be calculated on the basis of average of circularity measure C. Let
where n, is the number of correctly detected cells and c_k is the circularity measure of the k^th (k ε 1, 2, 3, ……. , n) correctly detected cell The higher the value of this parameter, the better the segmentation performance. The participants (suppose there are X participants) would be ranked on the basis of this parameter and the ranks [1 X] would be linearly mapped onto [20 5] which will be the value of the RD parameter.
The submitted report must contain description of the methods applied and a flow chart explaining each part. These will be reviewed by expert committee and review report will be sent to the authors. The weightage of this Report is 30 %.
The final ranks will be based upon the sum of all the four ranking parameteres
- All segmented results must be binary images as in the sample.png image
- All the participants must save segmented results of images in .png format.
- The segmented results must be named as 1.png, 2.png …etc.
- All result must be sent to email@example.com