Potential of textural indices for characterizing the spatial organization of tumor cells in radionuclide imaging

摘要: 427 Objectives Characterizing tumor heterogeneity using texture analysis in PET has shown promise to predict treatment response or patient survival oncology. Yet, the relationship between indices (TI) and biological needs be clarified. As a first step towards understanding meaning of TI, we studied whether TI measured from autoradiography (AR) correlated with spatial distribution cells as seen pathology slides. Methods Thirty days after implantation mammary derived transgenic MMTV-PyMT mouse injection 7.4 MBq FDG, was sacrificed. Tumor then sliced, AR performed slices were stained hematoxylin eosin. Using one slice including great variety cell density, defined 80 subregions 30 x pixels on images (pixel: 50x50 µm2) copied them image component which nuclei are colored blue-purple. We selected out displaying three patterns: A - large continuous map no more than small islands ( Results Homogeneity, Entropy, SRE LRE significantly different types pattern (>85% cells, called A+) composed majority extracellular matrix Conclusions can distinguish various density slides, paving way for image-based vivo histopathology appropriate tracers. How these results scale when replacing by micro is under investigation determine size clusters needed microPET provide useful information regarding distribution.