Dynamic contrast enhanced Magnetic Resonance Imaging (DCE-MRI) monitoring of locally Advanced Rectal Cancer using time Intensity Curve Shape Characterisation

Purpose: To investigate the relationship between DCE-MRI TIC-shape characterization and tumor response evaluation for locally advanced colorectal cancer (LARC). Methods and Materials: This study proposes normalization and piecewise linear fitting of TIC and a number of shape descriptors. This analysis was applied to the DCE-MRI examinations of 30 patients, all of whom were affected by LARC. Clinical TNM index (cTNM) was obtained before and after pCRT. After total surgical mesorectal excision (TME), pathological TNM (pTNM) was obtained from histological specimens. Two expert radiologists evaluated each DCE-MRI examination. TIC shape descriptors were computed for each ROI: adequate paired sample tests and Hotelling Trace Criterion were applied in order to identify the parameters best suited to indicate a positive response to the treatment. Results: In responders: Kruskal Wallis test showed that wash-in slope (WIS), wash-in intercept (WII) and the area under the wash-out phase (AUCWO) showed statistical differences (P<0.05) when this descriptors were calculated on the normalized curve; Wilcoxon test for paired revealed that WIS and AUCWO decreased after pCRT while WII showed a positive trend.HTC analysis best subset were WOS, AUCWO/ AUCWI according to pT stage (three classes: pT3, pT2, pT0-1, with J= 5391.9) and according to treatment response, WIS, AUCWI, AUCWO (two classes, responders / non-responders, J= 1.9*1017). Conclusion: Statistical tests between descriptors, before and after neo-adjuvant treatment, demonstrate that tumor response evaluation of LARC can be performed via DCE-MRI TIC shape analysis. Moreover, post-treatment DCE-MRI TIC shape descriptors allow for separation of patients in groups according to the pTNM tumor classification.