Quantitative characterization of preneoplastic progression using single‐cell computed tomography and three‐dimensional karyometry

The development of morphological biosignatures to precisely characterize preneoplastic progression necessitates high-resolution three-dimensional (3D) cell imagery and robust image processing algorithms. We report on the quantitative characterization of nuclear structure alterations associated with preneoplastic progression in human esophageal epithelial cells using single-cell optical tomography and fully automated 3D karyometry. We stained cultured cells with hematoxylin and generated 3D images of individual cells by mathematically reconstructing 500 projection images acquired using optical absorption tomographic imaging. For 3D karyometry, we developed novel, fully automated algorithms to robustly segment the cellular, nuclear, and subnuclear components in the acquired cell images, and computed 41 quantitative morphological descriptors from these segmented volumes. In addition, we developed algorithms to quantify the spatial distribution and texture of the nuclear DNA. We applied our methods to normal, metaplastic, and dysplastic human esophageal epithelial cell lines, analyzing 100 cells per line. The 3D karyometric descriptors elucidated quantitative differences in morphology and enabled robust discrimination between cell lines on the basis of extracted morphological features. The morphometric hallmarks of cancer progression such as increased nuclear size, elevated nuclear content, and anomalous chromatin texture and distribution correlated with this preneoplastic progression model, pointing to the clinical use of our method for early cancer detection.