Biography
Dr. Zev Leifer has an M.A. from Harvard University and a Ph.D. from New York University in Microbiology. He has been the course director of the Pathology Laboratory for over 30 years. Recently, he has published and spoken at international meetings in the area of pathology education. His work was written up in a featured section in the Springer monograph, “Digital Pathologyâ€. His current research specializes in the adaptation of commercial software for use in digital pathology education. A prime example is the work described here in the area of data mining.
Abstract
Diagnostic Pathology is a critical aspect in determining the nature of the disease process. Typically, a biopsy sample is converted to a slice of tissue on a glass slide, which is analyzed by a pathologist using a microscope. Today, we are in the world of digital pathology. The slide is digitized. The digital image is stored and retrievable. The image is viewed on a computer functioning as a microscope. Now, multiply this by hundreds or thousands of slides per day in a large medical center. Each slide may contain between 50 and 400 MB of data. In parallel, there is an obvious need to train medical students in the analysis of many pathologies, more so for the training of residents and the review of new or uncommon conditions by senior clinicians. Enter “Zev Leifer’s Lab†using Quartzy.com. This system, using a commercial product designed to track lab chemicals and supplies, has been adapted to deal with the data mining challenge of the massive storage of digital images. It is a listing of links to images stored in the collections of numerous medical institutions. The unique aspect lies in the metadata tags and the sorting capacity. One may search and organize by tissue type, pathology type, etc. This mined data of digital images can be used for study, testing and research.
Biography
Wolfgang Orthuber is orthodontist and mathematician and has interdisciplinary experience.
Abstract
It is well known that data representation on the web can be improved very much. There are a lot of proposals, but if we want maximal efficiency, there is not so much freedom. Maximal efficiency of the basal data structure is desirable to minimize costs. In this short contribution we want to recall http://arxiv.org/abs/1406.1065 which shows that on the web efficient and uniform definition of searchable information is possible using the basal data structurernURL (of online definition) plus sequence of numbers rnwhich is called \"Domain Vector\" (DV). At this the \"online definition\" defines in standardized (machine readable) way a \"Domain Space\" (DS) which is a metric space whose elements are the DVs. A DV can precisely represent every definable information, from a simple word to complex multidimensional information e.g. in science, medicine, industry. http://numericsearch.com shows a few examples and demonstrates searchability. The online definition can be multilingual, but the meaning of DVs is language independent. DVs are internationally uniform and comparable, they allow well defined similarity search. The users create the online definitions and with this the search criteria. The URL locates the definition and can be abbreviated. Existing online definitions can be reused in new definitions, so that search over multiple DSs is possible. One of the next steps is determination of the exact standard for DS definitions. Everyone who recognizes the potential of the above data structure and who wants to maximize efficiency of data representation on the web is invited to contribute.rn