High-resolution Dynamic PET for Breast Tumor Differentiation

Institution: University of California, Davis
Investigator(s): Guobao Wang, Ph.D. -
Award Cycle: 2015 (Cycle 21) Grant #: 21IB-0133 Award: $186,042
Award Type: IDEA
Research Priorities
Detection, Prognosis and Treatment>Innovative Treatment Modalities: search for a cure



Initial Award Abstract (2015)

Every year over 200,000 new patients are diagnosed with breast tumors in the United States. Not all the breast tumors are the same. Available data suggest that treatment for a patient should be tailored based on the molecular biology of individual breast tumor to prevent unnecessary or insufficient treatments. Accurate evaluation of the risk level of a breast tumor, meaning how aggressively the tumor grows and how likely it is to come back after treatment, is thus very important. These risk evaluations have a direct impact on the decision of physician and patients to choose the best treatment plan. However, there is currently no noninvasive clinical method to differentiate and stratify breast tumor risks. Molecular imaging with positron emission tomography (PET) can provide comprehensive biological characterization of breast tumors and may meet this unmet need in breast cancer.

†The question(s) or central hypotheses of the research: We hypothesize that high-resolution dynamic PET, which makes a clear movie of a breast tumor, can provide much more information about the tumor biology than the traditional static PET does (which only provides a picture of the tumor). The four-dimensional (4D: 3D space+1D time) imaging features extracted from high-resolution dynamic PET has the potential to better characterize a breast tumor and predict its risk level more accurately. Our long-term goal is to develop 4D PET signatures of breast cancer for risk prediction to enable physicians to choose the most suitable treatment options. This potential of 4D PET, however, cannot be realized by using current whole-body clinical PET scanners because of the limited spatial resolution and sensitivity. We therefore propose to develop an advanced high-resolution 4D dynamic PET imaging system for breast tumor differentiation and risk evaluation.

The general methodology: This project will build upon the high-resolution breast PET/CT scanner that has been developed at University of California, Davis. We propose novel data processing algorithms to develop the dynamic parametric imaging capability of this 3D PET scanner to enable high-resolution 4D PET imaging. We will perform patient studies and demonstrate the potential of the high-resolution 4D breast PET in predicting breast cancer risk. The imaging outcomes will be compared with pathology results. The advocate community is delighted at the potential to risk prediction based on the advanced high-resolution 4D PET imaging. We have had a roundtable meeting with breast cancer survivors from "Save Ourselves" represented by Dr. Cass Brown Capel. We plan to present our research progress in local chapter meetings of these support and advocacy groups and invite advocates to visit our lab, review results and have discussions on the progress in the future.

Innovative elements of the project: The utilization of accurate, high resolution, 4D molecular information for improving risk evaluation and differentiation of breast tumors is the key innovation of this project. Our method for deriving blood input function for dedicated breast PET parametric imaging is novel. It does not require invasive blood samples as required by other methods. The novel image reconstruction algorithm to be developed in this project can reduce the noise by a factor of 5 compared with traditional approaches. The resulting high-resolution 4D breast PET system by this project will be the first ever system that can perform high-resolution dynamic breast PET imaging and will have much higher precision and accuracy than what is currently available by 3D PET or dynamic imaging using clinical whole-body PET. This novel high-resolution 4D PET technique will offer a precise molecular imaging tool to enable noninvasive risk stratification for breast cancer treatment.




Final Report (2018)

Conventional positron emission tomograpny (PET) scans often examine three-dimensional (3D) spatial tracer distribution at a single time point and provide standardized uptake value (SUV) as a semi-quantitative measure. In comparison, fourdimensional (4D: 3D space + 1D time) PET imaging acquires 3D images at multiple time points and provides both spatial and temporal distributions of breast tumor. 4D PET can potentially be used for differentiating cancer molecular subtypes and risk prediction in heterogeneous disease to enable a noninvasive way for physicians to choose the most suitable treatment options. Clinical whole-body PET scanners, however, cannot realize this potential because of the limited spatial resolution and sensitivity. In this project, we aim to explore the 4D dynamic parametric imaging capability of an existing high-resolution 3D PET scanner built at University of California at Davis to enable high-resolution 4D PET imaging for breast tumor characterization.

We proposed to develop, implement and validate new data processing methods to enable parametric imaging on the breast PET scanner. In the project period, we have made significant progresses to achieve the goal and obtain preliminary data. First, based upon the kernel framework that the PI has developed for dynamic PET imaging, we developed two new types of kernels to improve dynamic PET image reconstruction. These new reconstruction approaches were validated using synthetic physical phantom evaluation for parametric imaging of breast tumor kinetics. We implemented and optimized the reconstruction algorithm on the dedicated breast PET scanner. Part of the results was published in the Proceeding of 2017 SPIE Medical Imaging. Second, we collected the dynamic PET data of ten patients on clinical whole-body PET scanner and discovered the strong correlation between the peak blood activity and the late-time blood activity in dynamic F18-fluorodeoxyglucose (FDG) PET studies. This new knowledge provides a direct basis for developing a patient-adaptive population-based input function method to allow tracer kinetic modeling on dedicated breast PET. The result has been presented in the 2017 Annual Meeting of Society of Nuclear Medicine and Molecular Imaging (SNMMI). Third, we optimized the rotation-based data acquisition protocol and modified the existing scanner table with several instrumental modifications to improve scan efficiency and patientís comfort. Fourth, the kernel method we have developed for PET reconstruction has also been extended for breast optical tomography reconstruction.

Besides the scientific progress, this project has also extended its impact by bringing new investigators into the field of breast cancer research. Both the PI and his postdoctoral fellow were new to the field prior to this CBCRP project and now have become engaged into the research of breast cancer imaging as evidenced by their new publications and awards during and after the project period.



Kernel-based anatomically-aided diffuse optical tomography
Periodical:Biomedical Physics & Engineering Express
Index Medicus: Biomed Physics Eng Express
Authors: Baikejiang R, Zhang W, Zhu D, Hernandez A, Shakeri S, Wang GB, Qi J, Boone J, Li C
Yr: 2017 Vol: 3 Nbr: 5 Abs: Pg:055022

Dynamic PET Image reconstruction for parametric imaging using the HYPR kernel method
Periodical:Proceedings of SPIE--the International Society for Optical Engineering
Index Medicus: Proc SPIE
Authors: BA Spencer, J Qi, RD Badawi, and GB Wang
Yr: 2017 Vol: 101324W Nbr: Abs: Pg:1-7