We aimed to develop a novel method of detecting changes in lung conditions during radiotherapy using temporal subtraction technique. Twenty patients who underwent radiotherapy were retrospectively assessed by calculating optimal direct similarity error (ODSE) between initial and mid-treatment registered images. Patients were grouped according to region in tumor size and atelectasis for lung of < 20 or ≥ 20 cm3, which analyzed two field regions (1024 × 768 pixels, 512 × 512 pixels). Correlations between ODSE and changes in lung conditions were analyzed based on effect of radiation dose; receiver operating characteristic (ROC) analysis was performed to evaluate whether changes can be detected during treatment period. The ODSE of 1024 × 768 pixels was changed to 1.00 (0.28–3.48) for lung lesion size of < 20 cm3 and 1.86 (0.55–6.58) for the ≥ 20 cm3 lung lesion size. ODSE of 512 × 512 pixels was 1.03 (0.40–2.12) for the region in tumor size and atelectasis of < 20 cm3 and 1.90 (0.39–27.8) for the ≥ 20 cm3 lung lesion size. The region under the curve values from ROC analysis were 0.796 (1024 × 768 pixels) and 0.983 (512 × 512 pixels). A novel method can visually and numerically help to detect changes in lung condition at early treatment stages. Using this method, difference between plan and actual positional relationship for target and risk organs that cannot be predicted at the time of planning can be avoided, ensuring high safety and accuracy in lung radiotherapy.
- Deformable image registration
- Lung cancer
- Temporal subtraction
ASJC Scopus subject areas
- Radiological and Ultrasound Technology
- Biomedical Engineering
- Radiology Nuclear Medicine and imaging