Top Banner
Smart Markers for Watershed- Based Cell Segmentation Can Fahrettin Koyuncu1, Salim Arslan1, Irem Durmaz2, Rengul Cetin- Atalay2, Cigdem Gunduz-Demir1* Present by: Xinyu Chang Email:[email protected]
12

Present by: Xinyu Chang Email:xchang@kent

Feb 22, 2016

Download

Documents

Elsa

Smart Markers for Watershed-Based Cell Segmentation Can Fahrettin Koyuncu1, Salim Arslan1, Irem Durmaz2, Rengul Cetin-Atalay2, Cigdem Gunduz-Demir1*. Present by: Xinyu Chang Email:[email protected]. Marker-Controlled Watershed Segmentation. - PowerPoint PPT Presentation
Welcome message from author
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
Transcript
Page 1: Present by: Xinyu Chang Email:xchang@kent

Smart Markers for Watershed-Based Cell Segmentation

Can Fahrettin Koyuncu1, Salim Arslan1, Irem Durmaz2, Rengul Cetin-Atalay2, Cigdem Gunduz-Demir1*

Present by: Xinyu ChangEmail:[email protected]

Page 2: Present by: Xinyu Chang Email:xchang@kent

Marker-Controlled Watershed Segmentation

The watershed transform is often applied to separating touching objects in an image problem.

The watershed transform finds "catchment basins" and "watershed ridge lines" in an image by treating it as a surface where light pixels are high and dark pixels are low.

The watershed transform works better if you can identify, or "mark," foreground objects and background locations

Page 3: Present by: Xinyu Chang Email:xchang@kent

(1) Bright pixels, pixels region with high intensity.(2) Dark-center pixels, pixels with low intensity

fully surrounded by bright pixels(3) Dark pixels, , pixels with low intensity partially

surrounded by bright pixels

3 types of pixels

Page 4: Present by: Xinyu Chang Email:xchang@kent

4 types of cell

(1) Bright pixels represent type I cells and boundary.(2) Dark-center pixels for type II cells(3) Dark pixels for both type III (darker none circular

cell )and type IV (apoptosis)

Page 5: Present by: Xinyu Chang Email:xchang@kent

Process chart

Page 6: Present by: Xinyu Chang Email:xchang@kent

Otsu thresholding with Sobel operator

In order to apply a Otsu thresholding to obtain all the foreground, yet the dark pixels would not get lost. With a Sobel operator The gradient has been take into consideration.

Dark pixels regions is defined as intensity smaller than the Threshold of Otsu and larger than k ・ tsobel

Page 7: Present by: Xinyu Chang Email:xchang@kent

Noise and artifacts elimination1. Eliminate narrow dark

pixel regions around the boundary.

2. Majority filter3. Fill holes under

certain area4. Erase dark pixel

region without any bright pixel

Page 8: Present by: Xinyu Chang Email:xchang@kent

Majority filter

The Majority Filter tool has two criteria to satisfy before a replacement can occur: the number of neighboring cells of a similar value must be large enough (Majority or Half), and those cells must be contiguous about the center of the filter kernel. The second criteria concerning the spatial connectivity of the cells minimizes the corruption of cellular spatial patterns.

Page 9: Present by: Xinyu Chang Email:xchang@kent

Smart marker extractionFor dark pixels and dark-center pixels

Page 10: Present by: Xinyu Chang Email:xchang@kent

Smart marker extractionFor Bright pixels

1. Dilate to get rid of the boundary2. A bright pixel cell has a circle area beyond a

threshold

Page 11: Present by: Xinyu Chang Email:xchang@kent

Result and comparison

Page 12: Present by: Xinyu Chang Email:xchang@kent

Thank you