Wavelets as chromatin texture descriptors for the automated identificationof neoplastic nuclei

Chromatin distribution reflects the organization of the DNA of a nucleus an d contains important cellular diagnostic and prognostic information. Feulge n staining of breast tissue enables the chromatin distribution of the nucle us to be visualized in the form of texture. Describing texture in an object ive and quantitative way by means of a set of texture parameters, combined with the study of the relationship of such parameters to the pathobiologica l cell properties, is useful both for reduction of the subjectivity inheren tly coupled to visual observation and for more accurate prognosis or diagno sis. We have presented an automated classification scheme for the diagnosis and grading of invasive breast cancer. The input to this scheme was a digitized microscopical image, from which nuclei were segmented. Chromatin texture w as described using a set of textural parameters that include first- and sec ond-order statistics of the image grey levels. The more recently developed wavelet energy parameters were also included in our study. Classification w as performed by a Knn-classifier, which is a versatile multivariate statist ical classification technique. We investigated the role of the tissue preparation technique and found that parameters derived from cytospins were better texture descriptors than tho se from sections. A 100% correct classification was achieved in a patient d iagnosis experiment and 82% in a nuclear grading experiment.