Feb. 17, 2012
PET/CT Delineates Tumours in Motion
A single static PET/CT scan has the potential to replace a 4D-CT scan when it comes to determining the internal target volume (ITV) of tumours affected by respiratory motion, according to researchers in the US. Writing in the Red Journal, the team compared both methods and concluded that the generated ITVs are similar enough to consider PET/CT a viable alternative to 4D-CT, with the added advantage that a single PET/CT scan could reduce the patient's X-ray exposure (Int. J. Radiat. Oncol. Biol. Phys. doi: 10.1016/j.ijrobp.2011.06.2002).
"To our knowledge, this is the first time that a single static PET/CT scan has been used to delineate the tumour ITV for lung/thoracic cancer patients," Osama Mawlawi, from the Department of Imaging Physics at the MD Anderson Cancer Center (Houston, TX), told medicalphysicsweb. "Previously, due to the patient's respiratory motion, a lung/thoracic lesion ITV had to be derived from a 4D-CT series that was characterized by a relatively high patient X-ray dose. A single static PET/CT scan could reduce the dose received by at least 60% compared with 4D-CT."
Determining the ITV
A critical part of treatment planning for any lung or thoracic tumour is building up an accurate representation of the tumour motion to avoid irradiation of healthy tissue. The approach outlined by Mawlawi and his collaborators at Rice University (Houston, TX) takes the observed PET image and assumes that it is a joint convolution of an ideal PET image, which is free from any motion or partial volume effects, and a motion blurring kernel (MBK).
"The clinician first draws a region-of-interest (ROI) on the PET image that encompasses the lesion," explained Guoping Chang, the lead author of the paper and a student under Mawlawi's supervision. "This ROI is then fed into an algorithm that we have written in Matlab, which automatically derives the tumour's MBK and its corresponding ITV. For a typical lung lesion that has a size of 2 to 20 cm3, the tumour's ITV can be derived within five minutes, although this is a post-processing task so the time to complete is not critical."
Phantoms and patients
Controlled motion was applied to two phantoms, each containing two spheres, in a proof-of-principle test of the team's joint convolution model. Subsequent analysis revealed that the two ITVs had an average similarity of 97.2±0.3%. Following this, eight patients referred for PET/CT and 4D-CT evaluation of non-small-cell lung cancer were studied.
Each patient underwent a whole-body CT scan, a whole-body PET scan and a 4D-CT scan. The joint convolution method generated the ITV using the whole-body PET and CT images, while a maximum intensity projection CT (MIP-CT) image was generated from the 4D-CT data in order to delineate the ITV. Once all data had been collated, the two ITVs were found to have an average similarity of 81±16.7%.
"If the 4D-CT and our technique captured the same tumour motion, we would not expect any major differences between the results," commented Mawlawi. "However, the reality is that the 4D-CT captures one cycle of the patient's respiratory motion while PET captures multiple respiratory cycles, which is a more reliable representation of the patient's respiratory motion. The ITVs generated by the different techniques had an average Dice coefficient of 0.8, suggesting that the two volumes were similar to one another."
Mawlawi and his colleagues are now carrying out further clinical studies. "We are also trying to improve our technique by incorporating models for a heterogeneous tumour uptake as well as a spatially variant PET point spread function, both of which represent real clinical settings more accurately," commented Mawlawi.
About the author
Jacqueline Hewett is a freelance science and technology journalist based in Bristol, UK.
© IOP Publishing Limited. Republished with permission from medicalphysicsweb.org
Imaging Physics Research
Basic Science Research Labs
- CCSG Small Animal Imaging Facility
- Computational Research Lab
- Digital Imaging Research Lab
- Kappadath Lab
- Magnetic Relaxometry Research Lab
- Magnetic Resonance Systems Lab
- Nuclear Imaging Research Lab
- Optical Imaging Lab: Biomedical Optics & NanoDiagnostics (BOND)
- Photoacoustic Imaging Research Lab
Translational Science Research Groups
- Advanced MR Neuro-Imaging Group
- Dynamic Imaging Group
- Image Guided Interventions & Therapy Group
- MR Applications Group
- Nuclear Medicine & PET Imaging Group