EVALUATION OF DIAGNOSES AND TUMOR VOLUMES USING FDG-PET, MRI, MAMMOGRAM AND PATHOLOGICAL TESTS IN BREAST CANCER

S. Jang¹, G.J. Grevera², L. Guan², G.J. A. Kassaee³, M.D. Schnall², and A. Alavi².

¹Department of Radiation Oncology, Mount Sinai Medical Center, New York, NY 10029 ²Department of Radiology and ³Department of Radiation Oncology, University of Pennsylvania Medical Center, Philadelphia, PA 19106

Objectives: The use of multimodality imaging is essential to the diagnosis and treatment of cancer. The purpose of this study was to evaluate the similarities and differences in diagnoses using FDG-PET, MR and mammogram breast imaging along with the pathological test, and to compare measured tumor volumes using FDG-PET and MR imaging for delineating target volume in radiation treatment planning.

Methods: FDG-imaging studies were performed on 12 female patients with suspicious breast lesions in order to evaluate the tumor characteristics observed by the accumulation and distribution of FDG in comparisons with MR, mammogram and pathological tests. Fully corrected FDG-PET images were used to define the probability of malignancy between 0 and 100% and metabolic tumor volume (MTV). The FDG-PET MTVs were measured using the voxels above 40% of the maximum count in the tumor region-of-interests (ROIs) drawn by a nuclear medicine physician. MR images were acquired using fat-saturated 3D spoiled gradient-echo (SPGR) technique with the patients in prone position. Post-gadolinium contrast T1-weighted images were reconstructed at 512x384 pixels. An MR radiologist measured the probability of malignancy and three dimensional lesion sizes. We developed a semi-deformable registration method, the pushpins, to evaluate the prone MR images in comparison with the supine FDG-PET images. The probabilities of malignancy were compared and tumor volumes are being compared between multimodality images.

Results: Pathological tests resulted in malignancies in 10 patients. MR and mammogram readings showed the sensitivity of 0.70 while PET readings demonstrated that of 0.80. In the probability of malignancy reading MR and mammogram yielded moderate correlation (0.6< r <0.7) with pathological results while PET produced a strong correlation (r=0.84). Strong correlations were observed between MR and mammogram (r=0.73), and between PET and mammogram (r=0.83). In general, FDG-PET MTVs were much smaller than prone MR tumor volumes.

Conclusion: The smaller FDG-PET MTVs compared to MR tumor volumes in most cases for breast tumors needs further investigation in order to provide complementary information for CT-based treatment planning in radiation therapy. FDG-PET indicated stronger correlations with both pathological results and mammogram than MR. This study demonstrated the usefulness of FDG-PET imaging for the diagnosis of breast cancer and suggested a possible contribution to the treatment planning for radiation therapy. The FDG-PET MTV and prone-MR lesion volume can be used separately or together using a deformable registration method in order to provide more comprehensive disease extent for CT-based treatment planning in the radiation therapy of breast cancer.