This project proposes to build an integrated software-based system for enhancing the accuracy of tumor change detection. The intent of the system, called AutoRegister, which is to be deployed on a clinical magnetic resonance imaging (MRI) scanner platform, is to automate the alignment of a patient's brain scan with that of a prior scan, such that subsequent offline tumor measurements do not have error introduced solely by differing slice orientation. While functionally similar to current auto-alignment technologies such as Siemens AutoAlign, the proposed technology is not sensitive to the inherent noise of such subject-nonspecific landmark- based techniques. Additionally the proposed technology is based upon a novel registration algorithm that is immune to outlier anatomy, such as a tumor, which often adversely affects techniques such as AutoAlign. In the United States, there are an estimated 13,000 deaths per year due to tumors in the primary central nervous system. Standard and experimental therapies rely on accurate measurement of tumor size change to assess treatment response and guide the course of a clinical trial. The project will build upon a foundation of existing technology developed at the Martinos Center for Biomedical Imaging at the Massachusetts General Hospital (MGH). It will make use of a novel 3D MR image registration algorithm designed by the co-PI, which is highly accurate within-subject and within-modality, and ignores voxels where no accurate match is possible, such as tumor tissue and surrounding partial-volume effects. Project collaborators include those with expertise in Siemens MRI scanner interfacing for slice prescription that developed the original AutoAlign tool. The firs aim of the project is to develop a laptop-based platform to connect to a Siemens MRI scanner console, able to retrieve a patient's prior scan, register (align) it with a custom 'scout'scan ru within-session, and send a 'slice prescription'to the scanner such that the subsequent within-session tumor-detection scans are aligned with the prior tumor-detection scan, transparently improving the accuracy of the downstream workflow of the neuro-oncologist or neuroradiologist.
The second aim of the project is to evaluate and validate the performance of the AutoRegister system. Collaborators conducting a separate upcoming clinical trial of a glioblastoma treatment will include the set of scans necessary to compare AutoRegister to AutoAlign and manual-alignment, in a within-session test-retest paradigm. Other collaborators will label the tumor change in each image set, for subsequent comparison.
The third aim i s to plan and establish the company processes necessary to meet FDA regulations covering the anticipated commercial product. In Phase II of the project, the technology will be ported to run directly on the scanner and re-validated on a larger dataset, such that the technology may be licensed or acquired by a scanner manufacturer. A Siemens representative has provided a Letter of Support of the proposed project, indicating their high level of interest in licensing or purchasing the proposed technology.

Public Health Relevance

In clinical practice, detecting change in the size of a brain tumor, which is critical in diagnosis and treatment, is still a difficult task for surgeons and oncologists. Accurate measurement of a tumor using magnetic resonance imaging (MRI) is adversely affected by differences in the position of the patient's head at the time of each scan, typically spaced weeks or months apart. The proposed technology, AutoRegister, greatly reduces this source of measurement variation with minimal change to the workflow of a neuroradiologist, allowing significantly more accurate tracking of tumor change than is currently possible.

National Institute of Health (NIH)
National Cancer Institute (NCI)
Small Business Technology Transfer (STTR) Grants - Phase I (R41)
Project #
Application #
Study Section
Special Emphasis Panel (ZRG1)
Program Officer
Narayanan, Deepa
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Corticometrics, LLC
United States
Zip Code
Wachinger, Christian; Golland, Polina; Kremen, William et al. (2015) BrainPrint: a discriminative characterization of brain morphology. Neuroimage 109:232-48