The most common procedure for prostate biopsy for cancer diagnosis and staging involves the introduction of small-gauge needles from the rectum under the guidance of trans-rectal ultrasound imaging (TRUS). In most cases TRUS differentiates poorly or not at all between normal tissue and cancer in the prostate and therefore provides minimal guidance for targeting localized regions at risk within the prostate. Our hypothesis is that this shortcoming can be overcome by methodology to deformably map a patient-specific atlas of suspicious tissue volumes within the prostate, built from pre-biopsy magnetic resonance (MR) and magnetic resonance spectroscopic (MRS) images, to TRUS images during biopsy to serve as targets for accurate needle placement. The near-term research described in this proposal involves pilot studies using a novel and especially effective class of statistically trainable deformable-shape model (SDSM) called m-reps to perform this mapping. The most common procedure for prostate biopsy for cancer diagnosis and staging involves the introduction of small-gauge needles from the rectum under the guidance of trans-rectal ultrasound imaging (TRUS). In most cases TRUS differentiates poorly or not at all between normal tissue and cancer in the prostate and therefore provides minimal guidance for targeting localized regions at risk within the prostate. The proposed methodology will overcome this shortcoming by mapping a patient-specific atlas of suspicious tissue volumes from pre-biopsy magnetic resonance (MR) and magnetic resonance spectroscopic (MRS) images to TRUS images during biopsy to serve as targets for accurate needle placement.
