ScatterNet for projection-based 4D cone-beam computed tomography intensity correction of lung cancer patients.

Background and purpose: In radiotherapy, dose calculations based on 4D cone beam CTs (4DCBCTs) require image intensity corrections. This retrospective study compared the dose calculation accuracy of a deep learning, projection-based scatter correction workflow (ScatterNet), to slower workflows: conventional 4D projection-based scatter correction (CBCT(cor)) and a deformable image registration (DIR)-based method (4DvCT). Materials and methods: For 26 lung cancer patients, planning CTs (pCTs), 4DCTs and CBCT projections were available. ScatterNet was trained with pairs of raw and corrected CBCT projections. Corrected projections from ScatterNet and the conventional workflow were reconstructed using MA-ROOSTER, yielding 4DCBCT(SN) and 4DCBCT(cor). The 4DvCT was generated by 4DCT to 4DCBCT DIR, as part of the 4DCBCT(cor) workflow. Robust intensity modulated proton therapy treatment plans were created on free-breathing pCTs. 4DCBCT(SN) was compared to 4DCBCT(cor) and the 4DvCT in terms of image quality and dose calculation accuracy (dose-volume-histogram parameters and 3%/3 mm gamma analysis). Results: 4DCBCT(SN) resulted in an average mean absolute error of 87 HU and 102 HU when compared to 4DCBCT(cor) and 4DvCT respectively. High agreement was observed in targets with median dose differences of 0.4 Gy (4DCBCT(SN)-4DCBCT(cor)) and 0.3 Gy (4DCBCT(SN)-4DvCT). The gamma analysis showed high average 3%/3 mm pass rates of 96% for both 4DCBCT(SN) vs. 4DCBCT(cor) and 4DCBCT(SN) vs. 4DvCT. Conclusions: Accurate 4D dose calculations are feasible for lung cancer patients using ScatterNet for 4DCBCT correction. Average scatter correction times could be reduced from 10 min (4DCBCT(cor)) to 3.9 s, showing the clinical suitability of the proposed deep learning-based method.