CBCT correction using a cycle-consistent generative adversarial network and unpaired training to enable photon and proton dose calculation

• Published in: Physics in medicine & biology Vol. 64; no. 22; p. 225004
• Main Authors: Kurz, Christopher, Maspero, Matteo, Savenije, Mark H F, Landry, Guillaume, Kamp, Florian, Pinto, Marco, Li, Minglun, Parodi, Katia, Belka, Claus, van den Berg, Cornelis A T
• Format: Journal Article
• Language: English
• Published: England 15-11-2019
• Subjects: Cone-Beam Computed Tomography; Deep Learning; Humans; Image Processing, Computer-Assisted - methods; Male; Photons; Proton Therapy; Radiation Dosage; Radiometry; Radiotherapy Dosage; Radiotherapy Planning, Computer-Assisted - methods; Radiotherapy, Intensity-Modulated
• ISSN: 1361-6560
• DOI: 10.1088/1361-6560/ab4d8c

In presence of inter-fractional anatomical changes, clinical benefits are anticipated from image-guided adaptive radiotherapy. Nowadays, cone-beam CT (CBCT) imaging is mostly utilized during pre-treatment imaging for position verification. Due to various artifacts, image quality is typically not sufficient for photon or proton dose calculation, thus demanding accurate CBCT correction, as potentially provided by deep learning techniques. This work aimed at investigating the feasibility of utilizing a cycle-consistent generative adversarial network (cycleGAN) for prostate CBCT correction using unpaired training. Thirty-three patients were included. The network was trained to translate uncorrected, original CBCT images (CBCT
) into planning CT equivalent images (CBCT
). HU accuracy was determined by comparison to a previously validated CBCT correction technique (CBCT
). Dosimetric accuracy was inferred for volumetric-modulated arc photon therapy (VMAT) and opposing single-field uniform dose (OSFUD) proton plans, optimized on CBCT
and recalculated on CBCT
. Single-sided SFUD proton plans were utilized to assess proton range accuracy. The mean HU error of CBCT
with respect to CBCT
decreased from 24 HU for CBCT
to  -6 HU. Dose calculation accuracy was high for VMAT, with average pass-rates of 100%/89% for a 2%/1% dose difference criterion. For proton OSFUD plans, the average pass-rate for a 2% dose difference criterion was 80%. Using a (2%, 2 mm) gamma criterion, the pass-rate was 96%. 93% of all analyzed SFUD profiles had a range agreement better than 3 mm. CBCT correction time was reduced from 6-10 min for CBCT
to 10 s for CBCT
. Our study demonstrated the feasibility of utilizing a cycleGAN for CBCT correction, achieving high dose calculation accuracy for VMAT. For proton therapy, further improvements may be required. Due to unpaired training, the approach does not rely on anatomically consistent training data or potentially inaccurate deformable image registration. The substantial speed-up for CBCT correction renders the method particularly interesting for adaptive radiotherapy.