Session Item

The Apparent Diffusion Coefficient (ADC) derived from diffusion-weighted imaging (DWI) in MRI for adaptative radiotherapy is a promising method to quantify early tumour response and thus to optimize dose distributions. We aim to compare several methods to obtain the ADC in cerebral structures and brain metastases.

MRI series were performed on a Philips 3T Ingenia magnet. Regions of interest were delineated with ARIA (Varian, v. 15.1) and ADC values were obtained with IntelliSpace Portal (Philips, v. 11.1) according to the monoexponential approach.

The mean ADC values of three brain structures (brainstem, ventricles and vitreous humour) from a cohort of four brain tumour patients were calculated from DWI using turbo spin-echo (TSE) (b = 0, 500, 1000 s/mm2) and echo-planar imaging (EPI), the latter with two phase-encoding directions (antero-posterior - AP, and postero-anterior - PA) and applying the reversed gradient (RG) algorithm to reduce geometric distortion. For EPI sequences, b was 0, 50, 100, 200, 500, 1000, 3000 s/mm², but we calculated ADC maps with 3 and 7 values (Fig.1).

Seven brain metastatic tumours were analysed from the four patients. Using dynamic studies (Ktrans), we differentiated between heterogeneously (n = 2) and homogeneously vascularised (n = 5) tumours, dividing the former into well and poorly vascularised areas, where mean ADC values were calculated separately.

ADC values from DWI-EPI with 7 b-values differ from those obtained with 3 b-values and DWI-TSE, which are closer to the expected ones (Fig. 1). This is due to the monoexponential approach, since lower and higher b-values are better adjusted to a biexponential curve. When distortion is not corrected, ADC values barely change for large structures (brainstem and ventricles), but it does for smaller ones (eyes), as wider dispersion reflects. RG correction reduces this dispersion. In previous works, we obtained similar results on a specific phantom made of water, ethanol and propanol.

Heterogeneously vascularised tumours showed lower ADC values in poorly vascularised areas, and higher in well-vascularised, similarly to small homogeneously vascularised tumours (Fig. 2). No significant difference was found between ADC values from TSE and EPI-RG (3b) except for small tumours, which can be explained by distortion.

Equivalent ADC values can be obtained with DWI-TSE or DWI-EPI if distortion is corrected (RG). The monoexponential approach requires few b-values and the biexponential must be used to consider more b-values. TSE implies longer acquisition time, but only one series is needed. Using EPI, a distortion correction algorithm must be used, which requires the acquisition of two series, lengthening the process. Well-vascularised areas of heterogeneous tumours and small tumours show similar ADC values, greater than poorly vascularised areas.

Funded by ISCIII PI17/01735 grant (co-funded by FEDER) and SINFONIA project (EURATOM 2019-2020 under GA No.945196).