Pattern recognition of microcirculation with super-resolution ultrasound imaging provides markers for early tumor response to anti-angiogenic therapy.

Cancer treatment outcome is traditionally evaluated by tumor volume change in clinics, while tumor microvascular heterogeneity reflecting tumor response has not been fully explored due to technical limitations. We introduce a new paradigm in super-resolution ultrasound imaging, termed pattern recognition of microcirculation (PARM), which identifies both hemodynamic and morphological patterns of tumor microcirculation hidden in spatio-temporal space trajectories of microbubbles. PARM demonstrates the ability to distinguish different local blood flow velocities separated by a distance of 24 μm. Compared with traditional vascular parameters, PARM-derived heterogeneity parameters prove to be more sensitive to microvascular changes following anti-angiogenic therapy. Particularly, PARM-identified "sentinel" microvasculature, exhibiting evident structural changes as early as 24 hours after treatment initiation, correlates significantly with subsequent tumor volume changes (|r| > 0.9, P < 0.05). This provides prognostic insight into tumor response much earlier than clinical criteria. The ability of PARM to noninvasively quantify tumor vascular heterogeneity at the microvascular level may shed new light on early-stage assessment of cancer therapy.