Julip Jung MS, Abstract Co-Author: Nothing to Disclose

Young Gi Kim BS, Abstract Co-Author: Nothing to Disclose

Helen Hong PhD, Presenter: Nothing to Disclose

Sung Il Hwang MD, Abstract Co-Author: Nothing to Disclose

Our method can be used to improve the performance of localization, detection and staging of prostate cancer in multi-parametric MR images.

Multi-parametric MR is increasingly mentioned in prostate cancer localization due to its possibility of combining anatomical information with functional information. T2wMR provides high spatial resolution and T1wMR provides good contrast for bleeding. In DWI, prostate cancer appears as high signal intensity while the ADC shows it as low signal intensity. Thus, we propose a multimodality image fusion on multiparametric MR images using signal correction, rigid and non-rigid registrations and color-coded mapping.

10 patients with prostate cancer were scanned using Philipse Achieva 3.0T TX MRI system. T2 Turbo Spin Echo images (TR/TE = 2500-3500/90-120 ms, flip angle = 90°, 512 x 512 matrix, slice thickness of 3 mm, 160 mm FOV), T1 Turbo Spin Echo images (TR/TE = 500-570/9 ms, flip angle = 90°, 512 x 512 matrix, slice thickness of 3 mm, 160 mm FOV) and DWI (TR/TE = 6500/79 ms, flip angle = 90°, 256 x 256 matrix, slice thickness of 3 mm, 240 mm FOV, water excitation with b value of 1000 s/mm2) were obtained. To correct the signal intensity of bleeding within the prostate in T2wMR, the signal intensity of hemorrhage area within prostate of T2wMR is substituted for that of T1wMR. To align the prostate on DWI to T2wMR, the transformation parameters of DWI are estimated by normalized mutual information-based rigid registration. To align the prostate on ADC map to T2wMR, the estimated transformation parameters are applied to ADC map. Then DWI and ADC map are color-coded and overlaid to T2wMR, respectively. To confirm prostate cancer localization, histopathology image is co-registered to T2wMR. For evaluation of our method, our result is visually compared with the location of prostate cancer indicated by a radiologist.

Our signal correction can differentiate prostate cancer from hemorrhage area in T2wMR. Our multimodality image fusion can effectively localize the prostate cancer by providing complementary information and easily compare with ground-truth of histopathology image.

Jung, J, Kim, Y, Hong, H, Hwang, S, Prostate Cancer Localization in Multi-parametric MR Images Using Multimodality Image Fusion.  Radiological Society of North America 2014 Scientific Assembly and Annual Meeting, - ,Chicago IL. http://archive.rsna.org/2014/14016793.html