An Automated Framework for Extraction of Nuclei and Cytoplasm of White Blood Cells from Peripheral Blood Smear Images

One of the most remarkable factors in blast cells diagnosis is the increment of nucleus-cytoplasm ratio. Some other parameters like size of the nucleus and the whole, and how round the nucleus is also affect the diagnosis.

In this study, the explicit characters of nucleus were used to identify it from the other parts of blood smear. The nuclei were thus extracted by k-means clustering. The cytoplasm was then extracted based on gray level histogram.

This study was conducted on 100 microscopic images. The method had a specificity of 82.5 and a sensitivity of 84 in the extraction of nuclei. The corresponding values for cytoplasm detection were 75.02 and 79.

In this Study present an automatic method to determine the nucleus and cytoplasm. Although the ratio of the cell nucleus to the cytoplasm is an important factor in determining the blast, but other signs such as detecting nucleolus are also effective in the final diagnosis.