Ex vivo study of prostate cancer localization using rolling mechanical imaging towards minimally invasive surgery

Locating and sizing tumor nodules in human prostate using instrumented probing - computational framework and experimental validation

Detection of tumor nodules is key to early cancer diagnosis. This study investigates the potential of using the mechanical data, acquired from probing the prostate for detecting the existence, and, more importantly, characterizing the size and depth, from the posterior surface, of the prostate cancer (PCa) nodules. A computational approach is developed to quantify the uncertainty of nodule detectability and is based on identifying stiffness anomalies in the profiles of point force measurements across transverse sections of the prostate. The capability of the proposed method was assessed firstly using a 'training' dataset of in silico models including PCa nodules with random size, depth and location, followed by a clinical feasibility study, involving experimental data from 13 ex vivo prostates from patients who had undergone radical prostatatectomy. Promising levels of sensitivity and specificity were obtained for detecting the PCa nodules in a total of 44 prostate sections. This study has shown that the proposed methods could be a useful complementary tool to exisiting diagnostic methods of PCa. The future study will involve implementing the proposed measurement and detection strategies in vivo, with the help of a miniturized medical device.

Identification of tumor nodule in soft tissue: An inverse finite-element framework based on mechanical characterization

Identification and characterization of nodules in soft tissue, including their size, shape, and location, provide a basis for tumor identification. This study proposes an inverse finite-element (FE) based computational framework, for characterizing the size of examined tissue sample and detecting the presence of embedded tumor nodules using instrumented palpation, without a priori anatomical knowledge. The inverse analysis was applied to a model system, the human prostate, and was based on the reaction forces which can be obtained by trans-rectal mechanical probing and those from an equivalent FE model, which was optimized iteratively, by minimizing an error function between the two cases, toward the target solution. The tumor nodule can be identified through its influence on the stress state of the prostate. The effectiveness of the proposed method was further verified using a realistic prostate model reconstructed from magnetic resonance (MR) images. The results show the proposed framework to be capable of characterizing the key geometrical indices of the prostate and identifying the presence of cancerous nodules. Therefore, it has potential, when combined with instrumented palpation, for primary diagnosis of prostate cancer, and, potentially, solid tumors in other types of soft tissue.