Precise tumour volume delineation is a major challenge in lung cancer radiotherapy (RT) because of respiration induced motion. Four-dimensional computed tomography (4DCT) using maximum intensity projection (MIP) dataset is commonly used method to generate internal target volume (ITV). This study compared ITV generation by tumour delineation on MIP and on all phases of the respiratory cycle.
Thirty consecutive patients who underwent 4DCT from Jan 2014 to Mar 2017 were included. After 4DCT image acquisition, ITV was generated by tumour delineation on MIP (ITVMIP) and on all ten phases of respiratory cycle(ITV10). The Correlation between ITVMIP and ITV10 was analysed using Matching index (M.I=overlapping/encompassing volume) and 3D centroid shift. M.I >0.8 was considered as good agreement. Time required to generate ITVMIP and ITV10 was also documented. Mann Whitney test was used for analysis and p < 0.05 was considered significant.
Mean volume of ITV10 was 134 cc (range 13 -627) and ITVMIP was 113 cc (range 11 - 569)(p < 0.001). Median volume of ITVMIP not getting covered by ITV10 was 4 cc (5.2%) and vice versa was 19.5cc (19%). The mean M.I between ITVMIP and ITV10 was 0.76 (SD = 0.08, range 0.57-0.88). For 15 patients with tumour close to high density structures like mediastinum, mean MI was 0.73 and for peripheral tumours was 0.78 (p = 0.003). Mean M.I was 0.75 for both tumour ≤5 and >5 cm. For peripheral and ≤ 5 cm tumours, M.I was 0.8. The average time for ITVMIP and ITV10 delineation was 9 and 96 minutes respectively. The centre of mass of ITV10 and ITVMIP was same 0.15 cm in sup/inf axis, 0.11 cm in medio/lateral axis and -0.06 and -0.04 cm in ant/post axis respectively when compared with free breathing CT. The 3D Centroid shift between ITVMIP and ITV10 was 0.01 cm.
ITV delineation using MIP is significantly smaller and may be imprecise in tumours adjacent to mediastinum, chest wall and diaphragm. This difference is likely due to difficulty in delineating tumour edges caused by blurring of images. However, delineation using MIP dataset is significantly less time consuming, hence, we recommend its continued use in peripheral tumour and advises caution in central tumours.
Clinical trial identification
Legal entity responsible for the study
All authors have declared no conflicts of interest.