|M.Sc Student||Ziv Omri|
|Subject||Noninvasive Microwave Thermal Ablation Monitoring Using|
X-Ray CT Image Features
|Department||Department of Biomedical Engineering||Supervisor||Professor Haim Azhari|
|Full Thesis text|
Efficient cancer treatment is a main challenge in today's medicine. Thermal ablation has become prevalent as an alternative treatment for surgery, since it is not always desired or possible to operate. One leading technology for tumor thermal ablation is using a needle type applicator that deploys at its tip energy in a form of microwave (MW) radiation. The MW energy leads to a substantial local temperature elevation and consequently to the destruction of the heated tissue region. After the thermal ablation treatment is completed, the body removes the coagulated tissue without the need for any additional surgical intervention.
Continuous monitoring is essential during the procedure in order to avoid cases of insufficient damage or oversized damage to the tissue. X-ray CT is currently used for guiding needle applicators. The aim of this study was to evaluate its potential use for thermal monitoring while considering the ionizing radiation involved.
In the ex-vivo part of the research, ten specimens of bovine liver were subjected to microwave ablation for the duration of 8 minutes using different power levels (20-80W). The specimens were scanned by X-ray CT at 5 seconds intervals during the process. Specimens were then cut along planes matching the imaging planes and the ablated zones were manually traced by two observers for reference.
Two algorithms for precise automatic mapping of the thermal ablation zone were developed. The first algorithm is based on Hounsfield Units (ISHU) changes and the second algorithm is based on radial optical flow (ROF). The quality of the automatic tracing produced by the algorithms was evaluated by were comparing them to the manual tracing. The sensitivity of the algorithms to spatiotemporal under-sampling was also evaluated in order to assess the feasibility of implementing them in radiation reduction scanning techniques. Both algorithms performed well, yielding radial accuracy <2mm in most cases.
For the in-vivo part, six microwave ablation procedures were performed in three pigs. In each procedure, a MW antenna was placed at a different location within the pig liver. The ROF algorithm was implemented in this case. The contrast enhanced images obtained post ablation were used as reference.
Three main challenges which have arisen during the transition from the ex-vivo experiment to the in-vivo. The first challenge was to synthetically generate a new set of X-ray CT slices which are perpendicular to the MW antenna. The second challenge was to correct the displacement between successive scans caused by breathing motion. The third was to avoid false tracing of the ablated area caused by vasodilation of proximate large blood vessels.
In conclusion, the results of both the ex-vivo and in-vivo experiments have shown that accurate non-invasive automated mapping of the ablation zone is feasible using X-ray CT with an acceptable radiation dose.