Predicting the target position accurately based on an external surrogate with quantifiable correlation is important for high-precision radiation in lung cancer. This study aimed to quantify the amount of movement of the lung tumours and understand the pattern of hysteresis based on four-dimensional (4D) computed tomography (CT) imaging compared to the chest wall movement.

From the radiotherapy four-dimensional computed tomography (4DCT) scan images, a total of 43 lung tumours from 21 patients were contoured, and movement in mediolateral (X), anteroposterior (Z), and superoinferior (Y) directions were calculated based on tumour centroid of the smart breath cine mode of the 4DCT scan. The tumour motion in different phases of the breathing cycle was calculated, and the deformation of the shape was illustrated using a 3D slicer. The nonlinear trajectory of the tumour motion resulting in tumour hysteresis was computed.

Tumour motion calculated from the 4DCT images in X, Z, and Y directions were 0.21±0.22 (range: 0.01–1.20), 0.18±0.15 (range: 0.01–0.49), 0.77±0.33 (range: 0.24–1.80) respectively. The mean three-dimensional radial motion vector was 0.85±0.31 (range: 0.21–1.81). No significant correlation between the magnitude of chest wall movement and three-dimensional radial vector was observed. Hysteresis in XZ plane was calculated to be 0.56±0.61cm² (range: 0.01–3.03). A statistically significant difference in hysteresis was observed between central and peripheral tumours (0.19±0.31cm² vs. 0.94±0.63cm², P<0.01).

4DCT-based tumour motion estimation is a feasible method, and the predictability of tumour motion by chest wall movement was not optimal. The movement was more for peripheral tumours compared to centrally located lesions. Location relative to midline was the most critical predictor of hysteresis. Considerable shape deformation in different phases of respiration was observed, and peripheral tumours had more than two times the motion during the breathing cycle compared to the central tumours.