Nonrigid registration method to assess reproducibility of breath-holding with ABC in lung cancer

To study the interfraction reproducibility of breath-holding using active breath control (ABC), and to develop computerized tools to evaluate three-dimensional (3D) intrathoracic motion in each patient. Since June 2002, 11 patients with non–small-cell lung cancer enrolled in a Phase II trial have undergone four CT scans: one during free-breathing (reference) and three using ABC. Patients left the room between breath-hold scans. The patient's breath was held at the same predefined phase of the breathing cycle (about 70% of the vital capacity) using the ABC device, then patients received 3D-conformal radiotherapy. Automated computerized tools for breath-hold CT scans were developed to analyze lung and tumor interfraction residual motions with 3D nonrigid registration. All patients but one were safely treated with ABC for 7 weeks. For 6 patients, the lung volume differences were 300 cm 3 and displacements >10 mm, probably owing to atelectasia and emphysema. One patient was excluded, and two others had incomplete data sets. Breath-holding with ABC was effective in 6 patients, and discrepancies were clinically accountable in 2. The proposed 3D nonrigid registration method allows for personalized evaluation of breath-holding reproducibility with ABC. It will be used to adapt the patient-specific internal margins.