Automated classification and enhanced characterization of circulating tumor cells by image cytometry

Enumeration and characterization of circulating tumor cells (CTC) is an emerging tool for the disease management of patients with metastatic carcinomas. CTC are correlated to progression free- and overall survival in several types of metastatic cancers, and can be used to predict therapy response. We have developed the CellTracks TDI system, which is an automated image cytometer that features continuous signal acquisition by a TDI camera to classify and characterize CTC in prepared blood samples from cancer patients. It measures quantitative fluorescence and morphological features at high sensitivity, high resolution, and is able to relocate events of interest for further investigation. An automated classification method, using Random Forests, was developed that eliminates the need for manual review of events and that classifies events as intact CTC, apoptotic CTC, CTC debris, leukocytes or debris, with a high agreement in classification between the automated classifier and 5 expert reviewers. Furthermore, we developed microstructures in PDMS that mechanically concentrate cells onto an analysis surface, thereby reducing the imaging- and image analysis times for a CTC sample. We also determined the quantitative and qualitative effect of free ferrofluid on imaging of cells in CTC samples that are processed by the CellTracks AutoPrep system. We found that the total fluorescence intensity of beads and cells is reduced by 11-67%, with the highest reductions in the blue region. An automated ferrofluid removal system was developed that removes 95% of the free ferrofluid in an AutoPrep sample with a recovery of CTC of 56-77%. In conclusion, the microscopic characterization and classification of CTC has been greatly improved by the results presented in this thesis. However, to serve as a liquid biopsy for the majority of patients, significant efforts to increase the percentage of patients in which CTC are detected are needed. Exhaustive characterization of all CTC in a patient will eventually lead to improved personal treatment, bringing cancer one step closer to being a chronic disease.