SBIR-STTR Award

A hardware-based Hybrid Lung Nodule Detection System will be developed for improving diagnostic accuracy and speed for lung cancerous pulmonary radiology. The configuration of the system includes the following processing phases: (1) data acquisition and pre-processing, in order to reduce and to enhance the figure-background contrast; (2) quick selection of nodule suspects based upon the most prominent feature of nodules, the disc shape; and (3) complete feature space determination and neural classification of nodules. Our R&D work will extend existing digital processing techniques, develop new ones, and introduce robust Artificial Neural Network (ANN) architectures in nodule detection and classification. We would like: (1) to automate the feature extraction and lung texture analysis techniques to improve the speed and accuracy of suspect nodule selection; (2) to analyze the detailed suspect nodules and to derive the additional relevant parameters and characteristic patterns, which are subsequently used for the classification task; and (3) to test and assess the developed hybrid (digital/neural) computation system's performance. A multistage ANN architecture involving an early supervised learning processing stage (e. g., a back propagation stage) followed by self-adaptive output stage (e. g., a Kohonen feature map stage), is currently being investigated to serve as a basis for testing the Hybrid Lung Nodule Detection System. This project will eventually lead to an efficient, cost-effective, robust, and hardware-based system for improving the accuracy and speed in nodule detection. It will also provide the basis for further development in other areas of diagnostic radiology.Awardee's statement of the potential commercial applications of the research:A hardware-based Hybrid Lung Nodule Detection System will be developed in order to assist pulmonary radiologists in detecting lung nodules. The detection system will enhance the current patient care system. The R&D will provide an effective tool to explore the integration of Artificial Intelligence into diagnostic radiology, especially the use of ANN to improve medical diagnostic technique in this and other cancer diagnostic procedures.National Cancer Institute (NCI)

An Expert-Driven Lung Nodule Detection (E-HLND) System is proposed for improving diagnostic accuracy and speed for lung cancerous pulmonary radiology. The research goal is to develop a robust, user-friendly, and clinically useful system to assist radiologists in the detection and analysis of lung tumor in an early and treatable stage. The detection and treatment of lung nodule in the early stage of growth can results in a better prognosis for survival. The proposed E-HLND system configuration include the following processing phases: (1) data acquisition of a large clinical screening chest x-ray films and multiresolution pre-processing to enhance object-to-background contrast, (2) quick selection of suspect nodule areas, (3) features space determination and neural classification of cancerous nodules as well as false positives, and (5) knowledge-based registration and fusion processing to integrate follow-up, patient history, radiologist's expertise, and other diagnosis reports. This proposal focuses on (l) improving system's sensitivity and specificity with neural network, image registration, information fusion technologies, and hardware design; and (2) validating system's performance with a "simulated" clinical trials based on a large clinical x-ray film database (Chinese Yunnan Tin Corp. Bio-Marker Specimen bank -YTC database). This project will explore artificial neural network and computer vision technologies in diagnostic radiology and provide a basis for other cancer research in diagnostic radiology. This R&D effort is not only consistent with but its success will provide good tool in the NCI launched large- scale study "Prostate, Lung, Colorectal, and Ovarian Cancer Screen (PLCO) Trial". PROPOSED COMMERCIAL APPLICATIONS An expert-driven lung nodule detection system, which serves as a "second reader" to assist pulmonary radiologists in detecting lung nodules, will be of great clinical and commercial value. The system can increase radiologists' sensitivity and specificity in the detection of early lung cancer on screening chest radiographs. Early and accurate detection of an early stage tumor will ensure patients get the best treatment available. The proposed R&D work will enhance current patient care system, reduce the work load of radiologists, and improve the cancer diagnostic procedure in diagnostic radiology.