REDUCING THE NUMBER OF PROSTATE BIOPSY CORES USING A REAL-TIME TOOL

Prostate cancer (PCa) is one of the most frequent neoplasia: it can be accurately detected only by prostate biopsy (PBx) and is not clearly visible at TRUS and other diagnostic methods such as PSA levels. For this reason, guidelines for PCa diagnosis suggest to perform a random tissue sampling when suspects arise from standard diagnostic methods. The sampling procedure is TRUS guided and roughly costs 20€/core, including the cost of the following histological analysis. Even if accuracy grows with cores’ number, costs, patient discomfort and adverse event probability grow as well. It is thus desirable to reduce the number of PBx cores without negative impact on diagnostic accuracy. This work describes the follow-up of a retrospective study aimed to the evaluation of the feasibility of a reduction in cores number through the application of a real time too. The studied technique enhances TRUS video streams with a live false color overlay image. Relative organ to transducer movements are tracked and compensated in order to provide a stable classification map which is useful to guide the physician during tissue sampling. The recorded signal is split into Regions of Interest (ROI) and statistical parameters are estimated with optimized algorithm. Finally, a nonlinear supervised classification model discriminates between PCa risk levels and results are superimposed to the TRUS image. The whole system was optimized for reducing the number of false positives while preserving an acceptable number of false negatives. Motion tracking improved the stability of the classification between frames. The proposed algorithm, thanks to its real-time image processing capabilities, may increase detection rate and could be helpful in finding region where cancer is represented.