Pilot Study On Consistency in Size Metrics for a Multimodality PEM/MR Breast Imaging Approach
M Sadinski1*, M Giger2, K Drukker3, K Yamaguchi4, L Lan5, H Li6, (1) Univ Chicago, Chicago, IL, (2) Univ Chicago, Chicago, IL, (3) Univ Chicago, Chicago , IL, (4) Univ Chicago, ,,(5) Univ Chicago, ,,(6) Univ Chicago, Chicago, ILTU-E-217BCD-7 Tuesday 2:00:00 PM - 3:50:00 PM Room: 217BCD
Purpose: In this study, we evaluated the degree of consistency between size metrics obtained from PEM and MRI to determine the intrinsic effectiveness of a multimodality approach using these two systems for breast imaging.
Methods: Under an IRB-approved protocol, 42 cases were considered (16 patients with 28 lesions), each consisting of an MRI data set and corresponding PEM data set, with an inclusion criterion of being obtained within 2 weeks of each other. Lesions were delineated manually on the PEM images and semi-automatically on the MRI images for efficiency. In addition to volume, equivalent sphere diameter (ESD) was evaluated for each lesion. This metric describes the diameter of a sphere with the same volume as that of the lesion, and is useful for minimizing the cubic impact of a single voxel contribution inherent to volume calculations.
Results: The relationship between PEM based volumes and MRI based volumes showed a linear trend around VolMRI=VolPEM demonstrating a degree of consistency in the two volumes. The correlation between VolMRI and VolPEM was calculated as 0.547 with a corresponding p-value of 0.00018 demonstrating a significant correlation. The associated concordance was calculated as 0.534. The ESD metric showed a more significant linear trend with correlation 0.733 and corresponding p-value 3.5x10-8, indicating consistency in lesion size. The concordance is 0.717, indicating high reproducibility.
Conclusion: Multimodality PEM/MRI breast imaging has the potential to combine functional and molecular imaging information for a powerful tool in cancer staging and evaluation of response to therapy. The resulting, expanded data set is of use only if fundamental size metrics are consistent between the two modalities. Our pilot data demonstrates that size metrics as we extracted from the image data are consistent between PEM and MRI breast image sets.