OA048] Analysis of diffusion and haemodynamic maps from multi-parametric MRI of prostate cancer

Abstract

Purpose Multi-parametric MRI (mp-MRI) is being introduced in radiotherapy (RT) of prostate cancer for tumour delineation in focal boosting strategies. It has been shown that the apparent diffusion coefficient (ADC) from diffusion weighted MRI (DW-MRI) may reflect cell density. Apart from tumour volume and cell densities, tumour hypoxia is also an important determinant of RT response. In this study we investigate the patterns of ADC and haemodynamic maps, and the relations between them, inside the index lesion. Methods ADC and perfusion maps from 10 prostate cancer patients were used, with the prostate and index lesion delineated by an experienced uro-radiologist. Associations between different ADC and perfusion histogram parameters inside the index lesion were evaluated with the Pearson’s correlation coefficient (PCC). In the voxel-wise analysis, scatter plots of ADC vs perfusion were analysed for voxels in the index lesion, again with the associations quantified with the PCC. Voxels that were in the lower quartile of both ADC and the forward transport rate constant ( K trans ) were assumed to be poorly oxygenated. Results ADC and the extravascular-extracellular volume fraction ( V e ) were overall lower inside the index lesion than outside. The opposite was observed for K trans that was higher inside the index lesion than outside. In the distribution analysis, the maximum K trans was significantly correlated with the maximum ADC (rho = 0.85). At the voxel level inside the index lesion, significant inverse correlation between ADC and K trans and significant positive correlation between ADC and V e were observed for nine of ten patients. According to our definition, 8% of the index lesion was poorly oxygenated. Conclusions A general negative correlation was seen in the index lesion between ADC and K trans while a general positive correlation was seen in the index lesion for ADC vs. V e . A method of estimating poor oxygenation in the index lesion was developed, based on diffusion and haemodynamic maps.