Residual tumour detection in post-treatment granulation tissue by using advanced diffusion models in head and neck squamous cell carcinoma patients.

Residual tumour detection in post-treatment granulation tissue by using advanced diffusion models in head and neck squamous cell carcinoma patients.

Eur J Radiol. 2017 May;90:14-19

Authors: Fujima N, Yoshida D, Sakashita T, Homma A, Kudo K, Shirato H

Abstract
PURPOSE: To evaluate the detectability of the residual tumour in post-treatment granulation tissue using parameters obtained with an advanced diffusion model in patients with head and neck squamous cell carcinoma (HNSCC) treated by chemoradiation therapy.
MATERIALS AND METHODS: We retrospectively evaluated 23 patients with HNSCC after the full course of chemoradiation therapy. The diffusion-weighted image (DWI) acquisition used single-shot spin-echo echo-planar imaging with 11 b-values (0-1000). We calculated 10 DWI parameters using a mono-exponential model, a bi-exponential model, a stretched exponential model (SEM), a diffusion kurtosis imaging (DKI) model and a statistical diffusion model (SDM) in the region of interest (ROI) placed on the post-treatment granulation tissue. The presence of residual tumour was determined by histological findings or clinical follow-up.
RESULTS: Among the 23 patients, seven patients were revealed to have residual tumour. The univariate analysis revealed significant differences in six parameters between the patients with and without residual tumour. From the receiver operating characteristic curve analysis, the highest area under curve was detected in the center of the Gaussian distribution of diffusion coefficient (Ds) obtained by the SDM. The multivariate analysis revealed that the Ds and diffusion heterogeneity (α) obtained by the SEM were predictors for the presence of residual tumour.
CONCLUSION: DWI parameters obtained by advanced fitting models will be one of the diagnostic tools for the detection of residual tumour.

PMID: 28583625 [PubMed - in process]

http://ift.tt/2sRJlS3