PathMiner: Web Telemicroscopy & Intelligent Databases

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Large-scale, multi-site collaboration is becoming indispensable for a wide range of research and clinical activities including tissue banking, proteomics, and outcome studies. Future progress in several key areas will rely on the capacity of individuals to dynamically acquire, share and assess images and correlated data. The main objective of this proposal is to develop, evaluate, and maintain PathMiner, a web-based model for interactive telemedicine, intelligent archiving, and automated decision support in pathology. The major components of PathMiner are a distributed telemicroscopy subsystem (DT), an intelligent archival (IA) subsystem and an Image Guided Decision Support (IGDS) subsystem. The DT subsystem enables users to control robotic microscopes, remotely, while the imaged specimen is broadcast to participants. It features entropy-based, auto-focusing, shared graphics, and messaging. The IA subsystem performs multivariate indexing and remote management of databases. The IGDS subsystem supports submission of queries across networks in order to automatically locate and retrieve digitized pathology specimens and correlated molecular studies of those cases from within distributed "ground-truth" databases which exhibit spectral and spatial profiles which are consistent with the query image. During preliminary studies, a prototype correctly discriminated among three commonly confused hematologic malignancies and normal cells in more than 83% of the cases using a mixed database of 1,900 test cells. The University of Medicine and Dentistry of NJ, Rutgers University, and the University of Pennsylvania School of Medicine will collaborate to 1) develop the means for populating ground-truth databases with new cases in multi-user environments; 2) expand ground-truth databases with a broader spectrum of hematologic disorders; 3) develop the means for unsupervised specimen analysis and determine the relationship between morphologic and clinical (protein/molecular) profiles; 4) determine confidence intervals in feature space for the spectrum of malignancies; 5) develop aweb interface for PazhMiner to coordinate flow of data and objects among subsystems; 6) conduct multi-site prospective performance studies and make PathMiner and the ground truth databases available as shareable resources for collaborative research, education and clinical decision support.