PROJECT SUMMARY/ABSTRACT An overarching goal of cancer screening is to detect cancer at an early stage while it is localized, treatable, and curable. However, cancer screening is associated with false positives, high rates of indeterminate findings, overdiagnosis, and overtreatment, which are serious limitations that need to be addressed to improve early detection efforts. Because medical imaging is a key component of early detection for many cancers, quantitative imaging/radiomics can provide biomarkers to address many of these limitations with early detection. Our group, Quantitative Imaging Clinical Validation Center at Moffitt Cancer Center (QICVC-MCC), helped pioneer image biomarker approaches leveraged in the prior funding cycle to create the first and only EDRN Clinical Validation Center (CVC) dedicated to the validation of image biomarkers. For breast cancer, we validated several breast density-type risk markers and diagnostic models in women classified as BI-RADS 4, noting the three subcategories within this classification were strong diagnostic markers, and constructed a bio-image repository for this subgroup. For lung cancer, we conducted extensive studies applying conventional radiomics for risk prediction, discrimination between malignant and benign nodules, distinguishing between indolent and aggressive lung cancers, predicting tumor mutations, and predicting treatment response. In this renewal, we will expand our CVC from validated conventional feature-based radiomics as a benchmark to compare end-to-end deep learning (DL) methods, expand to other populations, and implement AI platforms for analyzing breast, lung, and other organ site images. In breast imaging (Aim 1), we will expand our efforts from parametric modeling to machine learning/DL for improved risk, early detection, and diagnostic predictions and continue our data repository developments. In lung imaging (Aim 2), we will expand our efforts from lung cancer screening to incidentally detected nodules and surgically resected early-stage lung cancer. Additionally, in Aim 3 we will seek out additional opportunities within the EDRN to conduct studies of image biomarkers in other organ sites beyond breast and lung (e.g., prostate, pancreas, and cutaneous) to address emerging Network objectives. The EDRN has proven that it is greater than the sum of the individual projects. As such, in Aim 4 we propose to build a repository for the housing and sharing of images, algorithms, radiomics, clinical data, and information on biospecimens. In this CVC renewal, we will systematically validate radiomic features and novel image metrics in the early detection of cancer. This research is significant because such information may be able to complement existing clinical guidelines and lead to new strategies to apply noninvasive image biomarkers. The research of the QICVC-MCC is performed at an NCI-Designated Comprehensive Cancer Center, which is an outstanding environment to conduct such studies given the access to large patient populations and outstanding resources, and the clinical setting to deploy such biomarkers for improved personalized cancer care.

PROJECT NARRATIVE Cancer screening modalities are often associated with false positives, high rates of indeterminate findings, overdiagnosis, and overtreatment, which are all serious limitations that can be addressed with imaging. Our group pioneered image biomarker approaches leveraged in the prior funding cycle to create the first and only EDRN Clinical Validation Center (CVC) to dedicated to the validation of image biomarkers to improve early detection of cancer. In this renewal application, we build on our initial progress to bring practical automated solutions to these pressing screening problems.