A major focus of Dr. Foran’s laboratory is the development of a family of imaging and computational tools for characterizing a wide range of malignancies and elucidating associations and relationships among computational markers and gene, molecular and protein expression signatures throughout the course of disease onset and progression. The team has already reported the development and implementation of computational biomarkers that are generated by integrating histologic signatures and correlated genomic data using machine-learning and deep neural networks to predict disease recurrence in prostate cancer patients presenting with Gleason score 7 with greater accuracy than using PSA and standard clinical markers. These algorithms and methods will be modified and optimized for use in a range of other investigative and clinical applications.
The outbreak of the novel coronavirus (COVID-19) pandemic seriously endangered human health and life. Due to limited availability of test kits in developing countries throughout the world the need for auxiliary diagnostic approaches has become apparent. Recent research has shown radiography of COVID-19 patient, such as CT and X-ray, contains salient information about the COVID-19 virus and might be useful as an alternative diagnostic method. Chest X-ray (CXR) due to its quick imaging time, wide availability, low cost, and portability has gained much attention in this regard. In order to reduce intra- and inter-observer variability, during radiological assessment, computer-aided diagnostic tools are being developed to supplement clinical decision making and subsequent disease management. As part of a multi-institutional project the team has begun to explore the use of local weighted mean phase angle, phase energy and energy attenuation as the primary CXR image features that are assessed using multi-layered, convolutional neural networks. Quantitative evaluation was performed on data consisting of 8851 normal (healthy), 6045 pneumonia, and 3323 COVID-19 CXR scans resulting in 94.44% average accuracy across three classes of scans (COVID-19, Pneumonia, Normal). All COVID-19 cases were independently confirmed by reverse transcription polymerase chain reaction (RT-PCR). Although the results of these studies demonstrate the promise of these methods the team is expanding the experiments to test performance of the System in a much larger, open-set prospective investigation.
Another focus of the lab is the design, development and implementation of an enterprise-wide, clinical and research data warehouse that operates across Rutgers University and the RWJBarnabas Health system. The Warehouse automates the process of aggregating and organizing multi-modal data arising from Electronic Medical Records (EMR), Clinical Trial Management Systems (CTMS), Tumor Registries, Biospecimen Repositories, Radiology and Pathology archives and next-generation sequencing devices. The Warehouse is located in a secure high-performance computing environment that is equipped with machine-learning and analytic pipelines to support applications in precision medicine, data-mining, clinical outcome assessment and drug discovery.
Another ongoing effort in Dr. Foran's laboratory is focused on the emerging field of "radiomics." This discipline is aimed at extracting large amounts of quantitative features from medical images (computed tomography [CT], magnetic resonance [MR], positron emission tomography [PET], and/or ultrasound images) using algorithms and prediction models that can be used to determine the likelihood of an individual patient to benefit from a particular therapy.
Detailed description of individual projects can be found at: http://gemini.cinj.rutgers.edu/projects