Contrast-enhanced ultrasound (CEUS) involves the use of a microbubbles (MB) contrast agent and specialized ultrasound (US) imaging techniques. Any progress that can improve image quality and contrast, could positively impact patient management. Herein we introduce the concept of nonlinear filtering of backscattered US data for separating the tissue and MB signal components. Termed pulse inversion spectral deconvolution (PISD), this method does not require multiple pulse transmissions, which is typical for many nonlinear US imaging modes. Our new PISD-based CEUS approach uses two Gaussian-weighted Hermite polynomials (GH) to form two inverted pulse sequences. The two inverted pulses are then used to filter US backscattered data and discrimination of the linear and nonlinear signal components. A programmable US scanner (Vantage 256, Verasonics Inc) equipped with a linear array transducer was implemented with real-time PISD-based CEUS imaging. The receive data from all channels were shaped using plane wave imaging beamforming with angular compounding (1 to 9 angles). In vitro data was collected using a tissue-mimicking flow phantom perfused with a MB contrast agent (Definity, Lantheus Medical Imaging) using both PISD and traditional nonlinear US imaging as comparison. Contrast enhancement was quantified by computing the contrast-to-tissue ratio (CTR). Preliminary in vivo data was collected in the hindlimb of a healthy mouse. These in vitro and in vivo results collectively show that PISD-based CEUS imaging yields improved image contrast compared to the more traditional nonlinear US imaging approach.