Comparison of algorithms for IVIM modelling of DW-MRI with a clinical set of b-values for head and neck cancer: In silico and in vivo analysis.
Authors
Affiliations (4)
Affiliations (4)
- Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway. Electronic address: [email protected].
- Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway.
- Cancer Clinic, St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway.
- Department of Physics, Norwegian University of Science and Technology, Trondheim, Norway; Cancer Clinic, St. Olavs Hospital, Trondheim University Hospital, Trondheim, Norway.
Abstract
Intravoxel incoherent motion (IVIM) modelling of diffusion-weighted magnetic resonance imaging (DW-MRI) quantifies tissue diffusion (D<sub>t</sub>), capillary microcirculation (D<sub>p</sub>) and perfusion fraction (F<sub>p</sub>). However, accurate IVIM modelling using the few b-values acquired in clinical settings is challenging. This study mainly investigated the performance of four algorithms for accurate IVIM parameter modelling with few b-values before exploring the IVIM parameters' biomarker potential for head and neck cancer (HNC) radiotherapy response. Synthetic IVIM signals with 4, 5 and 11 b-values were generated. DW-MRI was acquired with 11 b-values for 10 HNC patients. Self-supervised (DNN<sub>SSL</sub>) and supervised (DNN<sub>SL</sub>) deep neural networks, and least square (LSQ) and segmented (SEG) fitting algorithms estimated IVIM parameters for both synthetic and patient imaging data using 4, 5 and 11 b-values. The algorithm accuracy was evaluated in silico and in vivo by calculating the error between the estimated and ground truth parameter values. A Kaplan-Meier analysis evaluated the associations between progression free survival (PFS) of 20 HNC patients and longitudinal changes in IVIM parameters estimated with 4 b-values. LSQ and DNN<sub>SL</sub> were the least and most accurate IVIM modelling algorithms, respectively. DNN<sub>SL</sub>, however, was biased towards the mean which limits the parameters' biomarker potential. None of the algorithms, however, estimated D<sub>p</sub> accurately with few b-values. Preliminary survival analysis suggested that an increase in F<sub>p</sub> during radiotherapy reduced the PFS of HNC patients. SEG and DNN<sub>SSL</sub> were the preferred algorithms for IVIM modelling few b-values, although reliable estimation was limited to the D<sub>t</sub> and F<sub>p</sub> parameters.