Dual-energy computed tomography improves delineation in primary brain-tumor patients

Purpose/Objective:

Dual-energy CT (DECT) improves the accuracy in proton therapy compared to single-energy CT (SECT). Since delineation of tumors and organs-at-risk (OARs) is gaining importance, we assessed whether DECT reduces the intra- and inter-observer delineation variability.
Material/Methods:
Two cohorts of 10 primary brain-tumor patients (adjuvant radio(chemo)therapy) each, receiving either 120kVp SECT or 80/140kVp DECT with identical dose, were evaluated. Four different pseudo-monoenergetic CT (MonoCT) datasets, representing several contrasts, were derived from DECT. Three radiation oncologists delineated the postoperative tumor bed volume (TBV) and several OARs. Delineations on SECT datasets were repeated once to assess intra-observer variability. Finally, delineations were performed on T1/T2-weighted MR scans as clinical reference.
The contour conformity was quantified by Jaccard index (JI) and Hausdorff distance (HD) between the contour intersection and union (Fig.1).
Results:
The median inter-observer TBV conformity (Fig.2A) was almost independent from CT acquisition (HD=6-9mm/JI=61-66%) and comparable to MR (HD=6-7mm/JI = 66-67%). The consistency of brainstem contours (Fig.2B) was best at the lowest energy (median HD=2.8mm/JI=81%). The conformity of parotid glands (Fig.2C) gained slightly from higher energies (0.6mm median HD reduction, 1% JI increase) and led to better results as MR. Smaller inter-observer variations were mostly achieved using the most suitable MonoCT instead of SECT.
The intra-observer TBV variability did not depend on clinical experience. However, less-experienced clinicians are more affected by different tissue contrasts (Fig.2D).
Conclusion:
For primary brain-tumor patients, DECT-derived MonoCT datasets improve intra- and inter-observer delineation conformity compared to SECT. Moreover, they in part led to similar or better results as the gold standard MR.