Automated nuclei detection in serous effusion cytology based on machine learning

Abstract

Serous effusions are common in clinical practice and they are frequently encountered specimen type in cytopathological assessment. Since this assessment is subjective, time-consuming and cause intra- and inter-observer variability, the need for an automated system is arised. Identification of the cancer cells in serous effusion cytology allows for the early diagnosis of the cancer and also the staging, prognosis and monitoring these cells. The detection of cell nuclei is seen as the corner stone for diagnostic purposes in automatic analysis of cytopathological images. Nuclei detection also yield the following automated microscopy applications, such as cell counting, segmentation and classification. In this paper, machine learning based Viola-Jones object detection approach is used to detect the cell nuclei locations in serous cytology images. When the method has been tested on number of serous cytology images, the obtained results show that this method has high nuclei detection performance.