Ultrasonic tissue-type imaging of prostate cancer

Prostate cancer (PCa) is diagnosed by means of needle biopsies typically guided by transrectalultrasound (TRUS). TRUS provides an inexpensive and effective means of visualizing gland anatomy and systematically directing each biopsy needle to a specific region within the gland. However, identification of cancerous regions in the prostate cannot be performed reliably using TRUS imaging or any other standard imaging modality. As a result, the majority of prostate biopsies are acquired blindly, with little or no knowledge of where cancerous tissue is located. Analysis of repeat-biopsy results for patients with initially negative biopsies indicates that half the cancers may be missed using standard biopsy-guidance imaging methods. Furthermore, because cancerous regions cannot be imaged reliably, region-specific focal treatment is not yet possible; all standard treatment procedures for gland-confined disease must be applied to the entire prostate. Prostate-tissue characterization based on a combination of ultrasound spectral (USS) parameters and clinical parameters that are classified by non-linear analysis methods, such as artificial neural networks (ANNs) and support-vector machines (SVMs), is showing encouraging potential for distinguishing cancerous from non-cancerous tissue in the prostate. These classification methods provide a means of generating tissue-type images (TTIs) that can reliably image cancerous regions in the prostate and provide information for guiding biopsies and planning and performing focal therapy for prostate-cancer treatment.