|
1
|
|
|
|
Evan D. Calabrese, MD, PhD
University of California, San Francisco
Bracco Diagnostics Inc./RSNA Research Resident Grant
(2020 - 2021)
MR Imaging Deep Learning and Radiomics Features for Guiding Individualized Therapy in Diffuse Gliomas
|
|
|
Abstract:
|
|
Diffuse gliomas are the most common primary brain malignancies, yet treatment options remain limited. Recent improvements in our understanding of diffuse glioma genetics have paved the way for numerous clinical trials of precision therapies targeted to specific tumor genetic biomarkers. Individualized tumor genetic assessment is rapidly becoming essential for accurate prognosis and for guiding emerging targeted therapies. Unfortunately, challenges remain for widespread tumor genetic testing due to costs and the need for tissue sampling. For these reasons, there is an increasing need for rapid, automated, non-invasive identification of specific diffuse glioma genetic biomarkers.
To address this need, our goal is to use multimodal MRI and artificial intelligence to predict clinically relevant diffuse glioma genetic biomarkers. Specifically, this study will focus on prediction of two common biomarkers that are being used to guide targeted therapies in ongoing clinical trials: EGFR amplification and CDKN2 loss. The feasibility of this approach is supported by our prior work demonstrating accurate, automated, MRI-based identification of IDH mutant gliomas. We hypothesize that a combination of deep learning and radiomics-based image features will allow accurate, automated, non-invasive prediction of EGFR amplification and CDKN2 loss in diffuse gliomas.
We plan to create a databased of over 500 diffuse glioma cases with multimodal MR imaging and 500 cancer gene sequencing results. This database will be used to develop an automated algorithm for predicting EGFR amplification and CDKN2 loss using a combination of deep learning and radiomics based MRI features. The results of work will facilitate the deployment of investigational targeted therapies to a broader patient population. In addition, the methods proposed here will be readily applicable for predicting many other clinically relevant genetic biomarkers. If supported, this proposal will be instrumental in helping Dr. Calabrese transition towards a career as an independent physician scientist in neuroradiology.
|
|
More Activities by Evan D. Calabrese, MD, PhD
|
|
|