Automated classification of cell morphology by coherence-controlled holographic microscopy

Abstract

Abstract. In the last few years, classification of cells by machine learning has become frequently used in biology. However, most of the approaches are based on morphometric (MO) features, which are not quantitative in terms of cell mass. This may result in poor classification accuracy. Here, we study the potential contribution of coherence-controlled holographic microscopy enabling quantitative phase imaging for the classification of cell morphologies. We compare our approach with the commonly used method based on MO features. We tested both classification approaches in an experiment with nutritionally deprived cancer tissue cells, while employing several supervised machine learning algorithms. Most of the classifiers provided higher performance when quantitative phase features were employed. Based on the results, it can be concluded that the quantitative phase features played an important role in improving the performance of the classification. The methodology could be valuable help in refining the monitoring of live cells in an automated fashion. We believe that coherencecontrolled holographic microscopy, as a tool for quantitative phase imaging, offers all preconditions for the accurate automated analysis of live cell behavior while enabling noninvasive label-free imaging with sufficient contrast and high-spatiotemporal phase sensitivity.

Machine Learning with Optical Phase Signatures for Phenotypic Profiling of Cell Lines

“…Whole-slide digital pathology (50) and a high-throughput well-plate culture imaging techniques (51,52) require robust, automated methods to identify cells of clinical interest. The classification accuracy from this study of 90-100% is high, similar to other reported values using machine-learning algorithms trained on optical phase map data (8,49,53). A leaveone-out classification analysis (Supporting Information Table S2) suggested predictive power derived from most of the 17 parameters, which could be reduced to five to six principal components, in accordance with other studies (2,8,49).…”

“…A leaveone-out classification analysis (Supporting Information Table S2) suggested predictive power derived from most of the 17 parameters, which could be reduced to five to six principal components, in accordance with other studies (2,8,49). Taken together, these results indicate that phase parameters from the pixel histogram and gray-level co-occurrence matrix, as well as cell outline-based morphology features (8,32,53), help to classify adherent cell lines. Combining "two-dimensional" and "three-dimensional" phase parameters into signatures input to machine learning makes the algorithm more flexible, as geometric/two-dimensional parameters are likely more important for classifying cell lines of dissimilar shape, and higher-order parameters more important for classifying cell lines of similar shape.…”

Quantitative assessment of cancer cell morphology and motility using telecentric digital holographic microscopy and machine learning

“…and ). The eleven quantitative parameters from this study overlap in part with morphological and phase parameters found to be useful in other DHM studies of cancer cells . Several other groups using holographic microscopy techniques have tracked migrating cancer cells , and identified phase textures of adherent cancer cells .…”

“…Numerous learning approaches are available and have been used on DHM data sets, including deep learning neural networks to distinguish SW480 colon cancer cells in suspension from white blood cells , anthrax from related microorganisms , and to detect and compensate for DHM background and phase aberrations . Machine learning paired with DHM has also recently been used to detect red blood cell infection by P. falciparum , discriminate between isogenic cell lines of differing metastatic stage , distinguish cancer grades from prostate biopsy microarrays , and distinguish healthy and nutrient‐deprived cancer cells , and live and dead yeast cells . Recent commentaries and reviews highlight recent studies combining these two techniques to advance cancer research .…”

“…A linear support vector machine learning algorithm was used based on excellent classification performance in previous studies using DHM data sets . Further, the method of PCA to reduce the dimensionality of features extracted for machine learning is well‐established in the literature .…”