Skip to Main Content

An official website of the United States government

Principal Investigator
Jaffar Ali Shahul Hameed
Florida Gulf Coast University
Position Title
Associate Professor
About this CDAS Project
NLST (Learn more about this study)
Project ID
Initial CDAS Request Approval
Oct 7, 2015
The Autonomous Collection and Databasing of Cancer and Structural Specific Features Present in CT Based Cancer Images through mathematical imaging based Systems
Modern Vision and Mathematical Imaging techniques have long been used in an attempt to gain information. We’ve seen this many times throughout cancer segmentation works. Often, this information requires human interference in an attempt to obtain meaningful or project specific information but loses out on global information present with the image. For example, if one is to segment solely with the goal of tumor identification, we often lose out on the quantitative features present within the environment of which the tumor is contained. It is with these losses we often negate meaningful features as means of project specific identifications. Such as in the case of tumors, we may segment the surrounding structures such as vessels and bones in an attempt to solely identify the tumor, thereby not collecting the maximal amount of information present within the image. The autonomous collection of information refers the ability of the system to collect and categorize many features present within the image(s) of interest without the intent of human interference among algorithm and processing selection. This would allow for a more brute force but autonomous method of retrieval among information collection techniques to gain the maximal amount of features in the image so as to be data mined with the intent of prediction based methods. The project aims at asking the question: How much data can we collect on cancer specific features present in the images and also within environment i.e. structural features such as vessels, bone structures, quantitative measurements, changes through time, and more to be used within large scale data mining of cancer images. The question poses the challenge of finding the maximal amount of feature present within the image, with least amount of human based interference possible. We attempt to build a mathematical system that can collect more information than modern techniques with a generalized autonomous process. With the intent to aid researchers with a system that can collect maximal amounts of cancer and structural specific features, without the need for researchers to solely build a system that over fits to the specific feature of interest.

Identification and collection of a maximal set of features present within CT and Pathology based images autonomously for the use of many researches without the need of human interference and system specific loss of features.


Brett Schwartz: Florida Gulf Coast University