MRI Registration for Therapy Evaluation and Annual Screening

Institution: University of California, Irvine
Investigator(s): Muqing Lin, M.S. -
Award Cycle: 2010 (Cycle 16) Grant #: 16GB-0056 Award: $76,000
Award Type: Dissertation Award
Research Priorities
Detection, Prognosis and Treatment>Imaging, Biomarkers, and Molecular Pathology: improving detection and diagnosis



Initial Award Abstract (2010)

Breast MRI has gradually evolved from a research tool to an important clinical imaging modality. To improve performance and reduce costs, computer-aided analysis tools are urgently needed. One example is to compare images taken from the same woman at different times. This is important to monitor breast density changes in high-risk women, or for monitoring the response of cancer patients who will receive neoadjuvant chemotherapy (NAC) to shrink tumors before surgery. Due to the highly deformable nature of breast, the MRIs taken at different times often present breast images of different shapes, so the comparison difficulty level is much higher than for mammograms. Unfortunately, none of the commercial breast MRI analysis tools provide such a comparison-image registration function.

Thus, this project will contribute in two areas of breast disease imaging evaluation; 1) NAC response and changes in breast density, and 2) development of computer-aided technology.

Our specific aims are to develop:
1) various registration algorithms for different situations where the object boundary and possibly some internal landmarks are available, and validate these algorithms using properly designed phantom (i.e., simulated breast tissue) studies.
2) 4-D (3 spatial dimensions plus time) breast MR registration for augmenting response evaluation in patients receiving NAC. The planned computer-aided tool will follow radiologistsí interpretation steps based on spatially co-registered images.
3) a computer-aided tool to detect the local changes of fibroglandular tissues in normal breasts.

The new image analysis tools will be applied to serial MRI studies performed from the same patients (NAC patients and normal volunteers) at different times to measure local changes of breast density. The registration quality of the breast will be first investigated, and when satisfactory, the registration of fibroglandular tissue will be performed to analyze the shrinkage and expansion of dense regions.

This research may benefit patients who choose to receive NAC, and the status for high-risk women who are recommended to receive annual breast MRI for screening.




Final Report (2012)

Breast density is a well-known biomarker of breast cancer. The volumetric change of breast density may indicate the cancer risk as well as monitor the response of cancer patients who receive neoadjuvant chemotherapy (NAC) before surgery. The goals of our project are: 1) develop image registration methods to recover the breast shape differences arising from compression and motion; 2) refine current image registration methods to preserve the volume of tumor bed during registration to ensure the accuracy of tumor regions; 3) quantitatively evaluate the local breast density change.

In the first project year, we developed a robust method to detect breast density in MRI and proposed a new method to evaluate the local volumetric change of breast density. For the final project year, we have been mainly working on two issues: 1) template-based automated breast segmentation in MRI, and 2) finite element modeling (FEM) based registration.

In general, breast density is detected based on the segmentation of breast so it is important to develop an automated and robust method to segment breast to ensure the accuracy of breast density segmentation. However, due to the large variety of breast MR scans (shapes, density patterns, etc.), it is difficult to develop a breast segmentation method that can be applied to all kinds of cases. We have proposed a new segmentation strategy which is to use chest region as the template for segmentation. The quantitative evaluation showed our proposed automated breast segmentation can accurately outline the boundary of breast on MRI. This work can be widely applied to different breast image analysis on MRI.

Due to the highly deformable nature of breast, the scans of the same patient taken at different times often present breast images of different shapes, which can be very challenging for quantitatively evaluating the intra-¨patient changes. In order to recover the breast shape differences arising from various deformation, we developed a FEM-based registration framework and implemented it into a graphical user interface (GUI) program. The initial experiments show that our proposed method has promising applications to various 3D breast image analysis.

For the future works after this grant period, we will focus on the following goals: 1) validate and improve our proposed FEM-based registration by testing on more breast MRI scans; 2) further improve our proposed automatic breast segmentation method and extend it to the application to breast MR fat-sat scans and breast CT scans; 3) implement an automatic landmark registration method based on SURF (speeded-up robust feature).



Spatial shrinkage/expansion patterns between breast density measured in two MRI scans evaluated by non-rigid registration
Periodical:Physics in Medicine and Biology
Index Medicus: Phys Med Biol
Authors: Lin M, Chen JH, Mehta RS, Bahri S, Chan S, Nalcioglu O, Su MY
Yr: 2011 Vol: 56 Nbr: 18 Abs: Pg:5865-75

A new bias field correction method combining N3 and FCM for improved segmentation of breast density on MRI.
Periodical:Medical Physics
Index Medicus: Med Phys
Authors: Lin M, Chan S, Chen JH, Chang D, Nie K, Chen ST, Lin CJ, Shih TC, Nalcioglu O, Su MY
Yr: 2011 Vol: 38 Nbr: 1 Abs: Pg:5-14

Consistency of breast density measured from the same women using different MR scanners.
Periodical:Annals of Oncology
Index Medicus: Ann Oncol
Authors: Chen JH, Chan S, Chang DH, Lin M, Su MY
Yr: 2011 Vol: 22 Nbr: 12 Abs: Pg:2693-4

Breast cancer: evaluation of response to neoadjuvant chemotherapy with 3.0-T MR imaging.
Periodical:Radiology
Index Medicus: Radiology
Authors: Chen JH, Bahri S, Mehta RS, Kuzucan A, Yu HJ, Carpenter PM, Feig SA, Lin M, et al.
Yr: 2011 Vol: 261 Nbr: 3 Abs: Pg:735-43