Abstract

In this paper we present two new methods of segmentation that we developed for nuclei and chromosomic probes – core ob jects for cytometry medical imaging. Our nucleic segmentation method is mathe- matically grounded on a novel parametric model of an image histogram, which accounts at the same time for the background noise, the nucleic textures and the nuclei’s alterations to the background. We adapted an Expectation-Maximisation algorithm to adjust this model to the his- tograms of each image and subregion, in a coarse-to-fine approach. The probe segmentation uses a new dome-detection algorithm, insensitive to background and foreground noise, which detects probes of any intensity. We detail our two segmentation methods and our EM algorithm, and dis- cuss the strengths of our techniques compared with state-of-the-art ap- proaches. Both our segmentation methods are unsupervised, automatic, and require no training nor tuning: as a result, they are directly applica- ble to a wide range of medical images. We have used them as part of a large-scale pro ject for the improvement of prenatal diagnostic of genetic diseases, and tested them on more than 2,100 images with nearly 14,000 nuclei. We report 99.3% accuracy for each of our segmentation methods, with a robustness to di?erent laboratory conditions unreported before.