Automatic, Rapid, High-Resolution Metabolic Imaging for Improved Management of Patients with Glioma
Abstract
Why This Research Is Needed: Primary brain tumors are typically aggressive lesions that are difficult to treat and have a relatively poor prognosis for many patients. Although magnetic resonance imaging (MRI) is commonly used in the management of patients diagnosed with brain cancer, the structural MR images commonly obtained for clinical diagnosis do not provide a clear definition of the boundaries of these tumors because of their infiltrative nature, where tumor cells often hide within the surrounding healthy-appearing tissue. Current practices for treatment, typically surgery followed by radiation therapy, are left to rely on defining the extent of the cancer based purely on what can be seen on structural images, reflecting damage to blood vessels. As a result, surgeons have a hard time determining when a maximum safe resection of the tumor has been achieved, and radiation oncologists target the lesion seen on structural MRIs, often undertreating areas of hidden tumor cells, while at the same time overtreating potentially healthy brain tissue. New ways to define the extent of tumor cells in healthy brain tissue are urgently needed for guiding surgeons and radiation oncologists for improving patient care. As metabolic changes often precede anatomic and microstructural changes, employing noninvasive metabolic MRI methods can potentially enable earlier intervention for treatment modification. Our previous studies have integrated metabolic MRI within standard clinical MRI examinations and generated metrics that were able to differentiate tumors from the normal brain tissue, assess metabolism specific to certain tumor subtypes, and predict survival. These metabolic parameters can reduce ambiguities in interpreting changes observed on conventional anatomic images and hence, aid in the definition of response to therapy. Despite these benefits, the imaging of brain metabolism with MRI has yet to be adopted into routine clinical practice due to long scan times and complicated acquisition schemes that require a high level of expertise to attain good quality data, lower spatial resolution compared to other MR imaging, and a lack of automated processing and quantification pipelines to generate images of brain metabolism rapidly on scanners. Aims of Proposed Research: In this proposal, we aim to automate the entire process of acquiring and generating accurate, high-resolution images of brain and tumor metabolism. This noninvasive, quantitative tool will be incorporated into routine clinical workflow to aid clinicians in precisely identifying tumor boundaries using a special type of MRI that can image tumor metabolism combined with artificial intelligence (AI) to automate the process. The first aim will develop strategies for rapidly scanning and generating high-resolution metabolic images of the whole brain using customized AI-based algorithms to define the optimal scan plane and orientation by automatically finding the general location of tumor within the brain and regions outside the brain that need to be avoided or suppressed. Aim 2 will develop a fully automated post-processing workflow to enable spectral processing, quantification, and quality control at the push of a button. These AI-based pipelines will then be installed on the scanner console to automatically generate accurate, high-resolution metabolic maps in only a few minutes. Aim 3 will then evaluate the final tools in patients with recurrent glioblastoma who are about to receive surgery in order to determine their impact on patient management and predicting progression free survival when combined with other routinely acquired clinical MRI. Impact: The result will be a collection of whole-brain, super high resolution metabolic images that represent maps of tumor activity or aggressiveness at a resolution similar to that of standard structural MRIs (< 3mm^3). Acquiring the data in less than 10 minutes will allow it to easily be added to any existing
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Jan 04, 2024
- Source ID
- HT94252310510
Entities
People
- Yan Li
Organizations
- United States Army
- University of California, San Francisco