Abstract
We aim to improve existing prostate tissue type imaging methods to better distinguish between cancerous and noncancerous prostate tissue. In doing so, we hope to increase the effectiveness of biopsy guidance, therapy targeting, and treatment monitoring. Spectral parameters obtained from radiofrequency (RF) ultrasonic (US) echo signals acquired from biopsy regions, along with the PSA level were used to train a neural network classifier. This method produced an ROC curve area of 0.84 compared to 0.64 obtained from Bmode, image-based classification. In order to further improve our prostate tissue typing methods, the integration of ultrasonic and magnetic resonance (MR) methods, to take advantage of the independent information provided by US and MR is being investigated. We are developing effective means of 3D spatial coregistration of US and MR data along with histological data used for validation.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have