INTRA-TUMORAL METABOLIC HETEROGENEITY OF CERVICAL CANCER Perry Grigsby Tumor Heterogeneity Definition Pre-Treatment 600 500 Y Variables 400 300 200 100 0 -100 0 10 20 30 40 50 60 70 Percent Threshold 80 90 100 110 Heterogeniety = d(V)/d(T) Chemo-IMRT Timeline 7-8 weeks (~55 days) 4 days IMRT 4 days IMRT 4 days IMRT 4 days IMRT 4 days IMRT 4 days IMRT IMRT 1 day HDR Pre-tx FDG-PET & MR Plan IMRT 1 day HDR 1 day HDR 1 day HDR 1 day HDR 1 day HDR ADC-MR to guide brachytherapy Post-tx FDG-PET & MR Evaluate response Change in Heterogeneity During Treatment Cell Point Chart Error Bars: ± 1 Standard Error(s) -.2 Heterogeneity (dV/dT) -.3 -.4 -.5 -.6 -.7 -.8 -.9 -1 Pre-Treatment Week 2 Week 4 Cu-ATSM Hypoxia Axial A B C Coronal Sagittal Fraction of 64Cu-ATSM Threshold Volume Within the FDG Defined Tumor Volume Threshold Mean ± SEM 40% 0.773 ± 0.013 50% 0.536 ± 0.023 60% 0.357 ± 0.024 70% 0.202 ± 0.018 80% 0.087 ± 0.010 Progression Free Survival 64Cu-ATSM Heterogeneity 1 0.8 p=0.04 0.6 Hetero > -1.7 Hetero ≤ -1.7 0.4 0.2 0 0 5 10 15 20 25 30 Time (Months) 35 40 45 Non-Spatial Other Measures of Heterogeneity Texture & Shape of Metabolism A toy example for demonstration of constructing the co-occurrence matrix. (a) Test image of size 4x4 and 4 intensity levels (M=4). The corresponding co-occurrence matrix assuming single pixel distance in: (b) East direction and (c) all directions. The matrix is converted into a probability mass function by normalization so that the sum is equal to unity. Co-occurrence matrix for 40% SUV 0.1 0.05 0 8 6 4 2 5 6 7 8 Co-occurrence matrix for 40% SUV 0.05 0.04 0.03 0.02 0.01 0 8 6 8 6 4 4 2 2 0 0 Texture Features • • • • Energy Contrast Local Homogeneity Entropy Shape-based Features • • • • Eccentricity Euler Number Solidity Extent Volumetric run length type texture Feature Definition Short Run Emphasis SRE Long Run Emphasis Low Gray-Level Run Emphasis LRE Description M 1 nr i 1 P (i , j ) j2 j 1 Measures the distribution of short runs. The SRE is highly dependent on the occurrence of short runs and is expected large for fine textures. M N Measures distribution of long runs. The LRE is highly dependent on the occurrence of long runs and is expected large for coarse structural textures. 1 nr P(i, j ) j 2 i 1 j 1 1 nr LGRE N M N P (i , j ) i2 j 1 i 1 N 1 P (i , j ) i 2 nr i 1 j 1 Measures the distribution of low gray level values. The LGRE is expected large for the image with low gray level values. M High Gray-Level Run Emphasis HGRE Short Run Low Gray-Level Emphasis SRLGE M N Measures the distribution of high gray level values. The HGRE is expected large for the image with high gray level values. i P (i , j ) 2 j2 j 1 Measures the joint distribution of short runs and low gray level values. The SRLGE is expected large for the image with many short runs and lower gray level values. Short Run High Gray-Level Emphasis 1 SRHGE nr P (i , j ) i 2 j2 i 1 j 1 Measures the joint distribution of short runs and high gray level values. The SRHGE is expected large for the image with many short runs and high gray level values. Long Run Low Gray-Level Emphasis 1 LRLGE nr P (i , j ) j 2 i2 i 1 j 1 Measures the joint distribution of long runs and low gray level values. The LRLGE is expected large for the image with many long runs and low gray level values. Long Run High Gray-Level Emphasis LRHGE Gray-Level Non-uniformity 1 MN GLNU P(i , j ) nr i 1 j 1 Run Length Non-uniformity RLNU Run Percentage RP 1 nr 1 nr i 1 M M N N 1 M N P (i , j ) i 2 j 2 nr i 1 j 1 2 M P (i , j ) j 1 i 1 nr M N N Measures the joint distribution of long runs and high gray level values . The LRHGE is expected large for images with many long runs and high gray level values. Measures the similarity of gray level values through out the image. The GLNU is expected small if the gray level values are alike through out the image. 2 Measures the similarity of the length of runs through out the image. The RLNU is expected small if the run lengths are alike through out the image. Measures the homogeneity and the distribution of runs of an image in a specific direction. The RP is the largest when the length of runs is 1 for all gray levels in specific direction. 30 Volumetric run length type texture 31 Volumetric zone length type texture Feature Short Zones Emphasis (SZE) Long Zones Emphasis (LZE) Low Grey-Level Zone Emphasis (LGZE) Definition SZE LZE M 1 nr 1 nr LGZE N i 1 Q(i , j ) j2 j 1 M N Q(i, j ) j 2 i 1 j 1 1 nr M N Q(i , j ) i2 j 1 i 1 M N 1 Q(i , j ) i 2 nr i 1 j 1 High Grey-Level Zone Emphasis (HGZE) HGZE Short Zone Low Gray-Level Emphasis (SZLGE) SZLGE 1 nr M N Q(i , j ) 2 j2 j 1 i i 1 Short Zone High Gray-Level Emphasis (SZHGE) 1 SZHGE nr Q(i , j ) i 2 j2 i 1 j 1 Long Zone Low Gray-Level Emphasis (LZLGE) 1 LZLGE nr Q(i , j ) j 2 i2 i 1 j 1 Long Zone High Gray-Level Emphasis (LZHGE) LZHGE Gray Level Non-Uniformity (GLNU) 1 MN GLNU Q(i , j ) nr i 1 j 1 Zone Length Non-Uniformity (ZLNU) ZLNU Zone Percentage (ZP) ZP 1 nr M M N N 1 M N Q(i, j ) i 2 j 2 nr i 1 j 1 2 M Q(i , j ) j 1 i 1 N 2 nr M N 32 Volumetric zone length type texture 33 Chemo-IMRT Timeline 7-8 weeks (~55 days) 4 days IMRT 4 days IMRT 4 days IMRT 4 days IMRT 4 days IMRT 4 days IMRT IMRT 1 day HDR Pre-tx FDG-PET & MR Plan IMRT 1 day HDR 1 day HDR 1 day HDR 1 day HDR 1 day HDR ADC-MR to guide brachytherapy Post-tx FDG-PET & MR Evaluate response Diffusion Weighted Imaging T2W ADC Patient 2 Patient 1 Week 1 Olsen JR, et al. ASTRO 2011. Week 3 Week 5 Tumor Volume Comparison Tumor Sub-Volume Comparison Tumor Sub-Volume Comparison Diffusion Weighted Imaging ADC Segmentation DTI DTI Goals • Identify Biologic Targets • Improve Image Guided Therapy Improve Survival with IMRT/IGRT External RT & Brachytherapy 1 Cum. Survival .8 .6 .4 .2 0 0 20 40 60 80 100 Time 120 140 160 180 Decrease Complications with IMRT/IGRT External RT & Brachytherapy .25 Cum. Hazard .2 .15 .1 .05 0 0 20 40 60 80 100 Time 120 140 160 180 Future Directions Spatial Heterogeneity Pre-Treatment Intra-Treatment Post-Treatment Future Directions Whole-Body PET/MR Hybrid Imaging