Optimal filtering for improved perfusion sensitivity

Ultrasonic perfusion imaging requires high sensitivity to slow disorganized blood-flow patterns while simultaneously rejecting clutter and noise. We achieve high sensitivity to both vascular flow and perfusion without contrast enhancement by adopting the following sampling pattern. Doppler pulses sampled at 1 kHz were acquired to form a Doppler frame. Frames were recorded at 10 Hz over 10 seconds. We organized these echo data into a 3-D array along slow-time, frame-time, and spatial axes. The 3-D echo array was decomposed using a 3-D SVD method to effectively separate blood flow signals from clutter and noise. These techniques were applied to a murine melanoma model at 24 MHz to form power Doppler tumor images. The results show the vascular perfusion of lesions without contrast enhancement.