Learning cancer-related drug efficacy exploiting consensus in coordinated motility within cell clusters.

OBJECTIVE: The ability of cells to collectively move is essential in various biological contexts including cancer metastasis. In this work, we propose an automatic video analysis tool to correlate cell movement inhibition with replication block induced by dose-dependent chemotherapy administration. The novel approach combines individual and collective cell kinematic analysis performed over time-lapse microscopy video frames. Cells are first localized and tracked, and then kinematic descriptors are extracted for each track. Selective track identification is performed assuming diversified cell roles within the same cluster (spontaneously forming groups of cells), and finally individual results are grouped exploiting consensus of coordinated motility within cell clusters. Recognition performance of three different experimental conditions (no drug, 0.5-5 μM merged in the same condition, and 50μM) reached an average accuracy value of 88% over 958 different tracks collected in 36 clusters of diverse dimensions in eight independent experiments.

CONCLUSION: An extensive application of this methodology could give a different point of view of the cancer mechanisms.