To compare and evaluate intensity-based registration methods for computation of serial x-raymammogram correspondence.Methods:
X-raymammograms were simulated from MRIs of 20 women using finite element methods for modeling breast compressions and employing a MRI/x-ray appearance change model. The parameter configurations of three registration methods, affine, fluid, and free-form deformation (FFD), were optimized for registering x-raymammograms on these simulated images. Five mammography film readers independently identified landmarks (tumor, nipple, and usually two other normal features) on pairs of diagnostic and corresponding prediagnostic digitized images from 52 breast cancer cases. Landmarks were independently reidentified by each reader. Target registration errors were calculated to compare the three registration methods using the reader landmarks as a gold standard. Data were analyzed using multilevel methods.Results:
Between-reader variability varied with landmark and screen , with between-reader mean distance (mm) in point location on the diagnostic/prediagnostic images of 2.50 (95% CI 1.95, 3.15)/2.84 (2.24, 3.55) for nipples and 4.26 (3.43, 5.24)/4.76 (3.85, 5.84) for tumors. Registration accuracy was sensitive to the type of landmark and the amount of breast density. For dense breasts , the affine and fluid methods outperformed FFD. For breasts with lower density, the affine registration surpassed both fluid and FFD. Mean accuracy (mm) of the affine registration varied between 3.16 (95% CI 2.56, 3.90) for nipple points in breasts with density 20%–39% and 5.73 (4.80, 6.84) for tumor points in breasts with density .Conclusions:
Affine registration accuracy was comparable to that between independent film readers. More advanced two-dimensional nonrigid registration algorithms were incapable of increasing the accuracy of image alignment when compared to affine registration.
The authors are grateful to the MARIBS study (Ref. 28), Guy’s and St. Thomas’ NHS Foundation Trust, and Addenbrooke’s Hospital, Cambridge University Hospitals NHS Foundation Trust for providing the MRIs used to simulate mammograms in this study. S. Pinto Pereira and I. dos Santos Silva are part of the Cancer Research UK Epidemiology and Genetics Group. S. Pinto Pereira is funded by a Graduate Teaching Assistant scholarship from the London School of Hygiene and Tropical Medicine.
II. MATERIALS AND METHODS
II.A. Registration methods
II.B. Mammogram simulation
II.C. Algorithm training and testing strategy using simulated mammograms
II.D. Test set of real mammograms
II.E. Landmark identification in real mammogram test set
II.F. Statistical analysis of real mammogram landmark data
III.A. Algorithm training and testing using simulated mammograms
III.B. Description of cancer cases
III.C. Within-reader agreement
III.D. Between-reader agreement
III.E. Registrations versus readers’ performance
Data & Media loading...
Article metrics loading...
Full text loading...