Spatiotemporal methods for prostate cancer detection by contrast-ultrasound dispersion imaging

Prostate cancer is the most diagnosed form of cancer in men in western countries. The lack of reliable imaging solutions currently restricts treatment options to radical interventions. Reliable imaging of angiogenesis, which correlates with cancer aggressiveness, could enable efficient targeting of biopsies and focal therapies, but the complicated effects of angiogenesis on perfusion have prevented a breakthrough so far. As an alternative, contrast-ultrasound dispersion imaging (CUDI) has been proposed to obtain a better characterization of angiogenenic microvascular changes. CUDI makes use of dynamic contrast-enhanced ultrasound imaging after an intravenous contrast-agent bolus injection. Dispersion is estimated by spatiotemporal analysis of time-intensity curves (TICs) obtained at each pixel. Here we present an analytical framework for a dispersion analysis by estimation of the spatial TIC similarity. Data preprocessing is improved by a spatial Wiener deconvolution filter, which reduces the effects of the anisotropic spatial ultrasound resolution, and by TIC time windowing. In addition, we propose the TIC correlation coefficient as a new TIC similarity measure. A comparison of the resulting dispersion maps from 7 recordings with histology, obtained after radical prostatectomy, showed an increased spatial similarity in the presence of cancer in all patients. The cancer classification performance of CUDI was superior to all perfusion-related parameters and was improved by spatial filtering and windowing. Although an extended validation is required, these results confirm the promising value of CUDI for prostate cancer detection.