Optimization strategies for respiratory motion management in stereotactic body radiation therapy

Freislederer, Philipp (2020): Optimization strategies for respiratory motion management in stereotactic body radiation therapy. Dissertation, LMU München: Medizinische Fakultät

Abstract

Various challenges arise during the treatment of lung tumors with stereotactic body radiation therapy (SBRT), which is a form of hypofractionated high precision conformal radiation therapy delivered to small targets. The dose is applied in only a few fractions and respiratory organ and tumor motion is a source of uncertainty additional to interfractional set-up errors. Respiratory organ and tumor motion is highly patient-specific and it affects the whole radiotherapy treatment chain. In this thesis, motion management techniques for SBRT are evaluated and improved in a clinical setting. A clinical need for improvement has been present at the LMU university hospital for each issue addressed in this thesis: Initially, the usage of respiratory correlated computed tomography (4DCT), which is vital for SBRT treatment, was seen as impractical and prone to uncertainties in the data reconstruction in its current form. Therefore, the 4DCT reconstruction workflow has been improved to minimize these potential error sources. Secondly, treatment planning for tumors affected by respiratory motion was evaluated and subsequently improved. Finally, the treatment technique of respiratory gating was implemented at the clinic, which led to the need of evaluating the respiratory gating characteristics of the novel system configuration. At first, the 4DCT reconstruction workflow used in clinical practice was investigated, as in the presence of respiratory motion the knowledge of tumor position over time is essential in SBRT treatments. Using 4DCT, the full motion range of the individual tumor can be determined. However, certain 4DCT reconstruction methods can under- or overestimate tumor motion due to limitations in the data acquisition scheme and due to the incorrect sorting of certain X-ray computed tomography (CT) image slices into different respiratory phases. As the regular clinical workflow of cycle-based sorting (CBS) without maximum inspiration detection (and therefore no clear starting point for the individual breathing cycles) seemed to be affected by these potential errors, the usage of CBS with correct maximum detection and another sorting algorithm of the respiration states, so-called local amplitude-based sorting (LAS), both have been implemented for a reduction of image artifacts and improved 4DCT quality. The three phase binning algorithms have been investigated in a phantom study (using 10 different breathing waveforms) and in a patient study (with 10 different patients). The mis-representation of the tumor volume was reduced in both implemented sorting algorithms compared to the previously used CBS approach (without correct maximum detection) in the phantom and the patient study. The clinical recommendation was the use of CBS with improved maximum detection, as too many manual interventions would be needed for the LAS workflow. Secondly, a combination of the actual patient breathing trace during treatment, the log files generated by the linear accelerator (LINAC), and Monte Carlo (MC) four-dimensional (4D) dose calculations for each individual fraction was implemented as a 4D dose evaluation tool. This workflow was tested in a clinical environment for SBRT treatment planning on multiple CT datasets featuring: a native free-breathing 3DCT, an average intensity projection (AIP) as well as a maximum intensity projection (MIP), both obtained from the patient's 4DCT, and density overrides (DOs) in a 3DCT. This study has been carried out for 5 SBRT patients for three-dimensional conformal radiation therapy (3D-CRT) and volumetric modulated arc therapy (VMAT) treatment plans. The dose has been recalculated on each 4DCT breathing phase according the the patient's breathing waveform and accumulated to the gross tumor volume (GTV) at the end-of-exhale (EOE) breathing phase using deformable image registration. Even though the least differences in planned and recalculated dose were found for AIP and MIP treatment planning, the results indicate a strong dependency on individual tumor motion due to the variability of breathing motion in general, and on tumor size. The combination of the patient's individual breathing trace during each SBRT fraction with 4D MC dose calculation based on the LINAC log file information leads to a good approximation of actual dose delivery. Finally, in order to ensure precise and accurate treatment for respiratory gating techniques, the technical characteristics of the LINAC in combination with a breathing motion monitoring system as s surrogate for tumor motion have to be identified. The dose delivery accuracy and the latency of a surface imaging system in connection with a modern medical LINAC were investigated using a dynamic breathing motion phantom. The dosimetric evaluation has been carried out using a static 2D-diode array. The measurement of the dose difference between gated and ungated radiation delivery was found to be below 1% (for clinical relevant gating levels of about 30%). The beam-on latency, or time delay, determined using radiographic films was found to be up to 851 ms±100 ms. With these known parameters, an adjustment of the pre-selected gating level or the internal target volume (ITV) margins could be made. With the highly patient-specific character of respiratory motion, lung SBRT faces many additional challenges besides the specific issues addressed in this thesis. However, the findings of this thesis have improved clinical workflows at the Department of Radiation Oncology of the LMU University hospital. In a future perspective, a workflow using evaluation of the actual 4D dose in combination with accurate 4DCT image acquisition and specialized treatment delivery (such as respiratory gating) has the potential for a safe further reduction of treatment margins and increased sparing of organs-at-risk (OARs) in SBRT without compromising tumor dose targeting accuracy.

• Dokumententyp: Dissertationen (Dissertation, LMU München)
• Themengebiete: 600 Technik, Medizin, angewandte Wissenschaften ; 600 Technik, Medizin, angewandte Wissenschaften > 610 Medizin und Gesundheit
• Fakultäten: Medizinische Fakultät
• Sprache der Hochschulschrift: Englisch
• Datum der mündlichen Prüfung: 7. Oktober 2020
• 1. Bericht­erstatter:in: Parodi, Katia