Quantification of Tumor Heterogeneity with Molecular Images: Software Development, Optimization and Evaluation for Clinical Impact

Tumor heterogeneity has been known as an important biomarker for tumor aggressiveness and clinical outcome. The quantification of tumor heterogeneity with molecular images, by analyzing the local or global variation in the spatial arrangements of pixel intensity with texture analysis, possesses a great clinical potential for treatment planning and prognosis. However, the methodology for computing the tumor heterogeneity indices requires further investigation, evaluation and optimization before it can be applied as a clinical routine. In this grant proposal, we aim to investigate the usefulness of heterogeneity quantification with molecular images in three different directions. First, we will develop an open-source software package for image texture analysis and heterogeneity quantification. This software package will be made available to the investigators world-wide without any charge. Second, we will use digital phantoms to evaluate the requirement of image qualities for heterogeneity quantification, especially in the spatial resolution and noise levels. Finally, we will apply the developed technology to head and neck cancer images to evaluate the clinical usefulness. Specifically, we will evaluate whether heterogeneity quantification would improve the outcome prediction power for advanced head and neck cancer patients, as well as using texture analysis to diagnose the extra-capsular spread of lymph nodes from PET images. With the development and investigation of tumor heterogeneity analysis in this proposal, physicians and scientists will better understand the properties of tumor heterogeneity that can be used for cancer patient management.

Project ID:PB10301-0496
External Project ID:NSC102-2221-E006-301-MY2