Motion Pictures

When trying to gain an advantage over cancer, the more we know, the better. That’s why there’s always demand for better, more detailed imaging techniques that provide more information about what’s going on inside the patient’s body. But sometimes, simply increasing the resolution of the picture isn’t enough. Lung cancer is one such situation – even during imaging, patients have to breathe, and the motion of the lungs creates problems. What can we do to overcome the issues inherent in taking photographs of a moving organ? We first need to understand the limitations of what we already have.

Lung limitations

At the moment, treatment planning for lung cancer radiotherapy is performed using either static three-dimensional computed tomography (3D-CT) or moving four-dimensional (4D-CT) datasets that track the lungs continuously over time. The main objective of this pre-treatment phase is to plan the patient’s radiation therapy so that we can maximize dosage delivery to the tumor while minimizing dosage to the surrounding healthy lung tissue. But that planning is confounded by the patient’s need to breathe. That makes the tumor a “moving target” for radiation – meaning that static 3D-CT images just aren’t good enough, because the lack of information on respiration-induced motion could result in errors during treatment. Even using a 4D-CT dataset for planning has limitations because the technique doesn’t image continuous motion over a single respiration cycle; rather, it captures static snapshots at various points over multiple respiration cycles and combines them to yield information on an “average” respiration cycle. If the person being imaged can’t maintain a regular breathing pattern during image acquisition – not an uncommon problem in cancer patients – then the resulting motion information can be noisy and inaccurate.

Magnetic resonance imaging (MRI) could offer a reasonable solution, as it allows the image capture of continuous lung and tumor motion over multiple respiration cycles, thus overcoming the limitations of 4D-CT. So why don’t we use it regularly? Unfortunately, MRI doesn’t provide the photon density information required for radiotherapy planning – and their spatial resolution is significantly lower than in CT images. The interior of the lung appears black in MRI images and the airways within the lung can’t be seen clearly. These disadvantages haven’t made us give up on MRI techniques altogether, but their use in lung cancer radiotherapy planning still needs more research.

Inspired by cross-disciplinary discussions

Our work is actually the result of a collaborative project with a bioengineer, Poh Chueh Loo from Nanyang Technological University, and two clinical collaborators, Tan Cher Heng from Tan Tock Seng Hospital and Ivan Tham from the National University Cancer Institute. All three are interested in studying the effects of lung and tumor motion on radiotherapy treatment so, working together with them, we developed a new way of examining the lung. Our method allows us to study the motion of the lung and its interior structures – like airways or a tumor – in a continuous manner during breathing. Under the current standard of care, which involves using a single medical imaging modality, we can capture either the lung motion or the interior structures, but not both. So how do we do it? We fuse two modalities into a single technique to obtain all of the information at once.

For our technically inclined readers, the first imaging method we use is 4D-MRI, which lets us continuously track lung motion, but doesn’t provide any details about the lung interior. The second method is 3D-CT, which provides a high-resolution view of the internal lung structures, but only at a single point in time. What’s unique about our method? We use what we’ve termed “MRI+CT fusion” to mathematically combine the details in both sets of images, revealing information not visible using the current standard techniques (see Figure 1).
At the moment, radiotherapy planning requires the use of CT images; it can’t be done using MRI data alone. We hope that our MRI+CT fusion method will make it possible to use the motion information from MRI for treatment planning – and perhaps one day even replace standard CT scanning altogether, reducing patients’ radiation exposure and allowing tests to be repeated as often as necessary.

In the clinic?

Despite the clear benefits, it’s still a challenge to integrate our method into the existing radiotherapy clinical workflow – mainly because we foresee integration issues with the hardware and software hospitals currently use. The good news, though, is that these obstacles don’t seem to be dampening the clinicians’ enthusiasm. We’ve spoken to doctors working not just on the lung, but in other areas where this kind of information fusion will be useful – and it’s encouraging to see that they’re trying to apply our method to their clinical problems as well.

When we first started our research, our target audiences were the clinicians and radiation oncologists who specialized in treating cancer with radiotherapy – especially those who focused on lung cancer. But we soon realized that, without access to the hardware and software used in hospitals, it would be very difficult for us to convince the clinicians of the utility of MRI+CT fusion. So at this point, I’d say that the biggest obstacle to the application of our method is getting buy-in from the equipment manufacturers. Without their support, it will be very challenging to integrate our method into the clinical workflow. There are, of course, also a number of regulatory hurdles we have to clear before MRI+CT fusion can even be applied in a clinical setting.

Right now, we are in the midst of exploring new application domains. Our first study focused on the lungs and how their motion affects tumor imaging, but MRI+CT fusion can be extended to studying other organs, too. We’re talking to clinicians about other problems where combining information from two – or even more – imaging techniques might be useful. Hopefully, we’ll soon be able to expand to other applications. It would be nice to see our method becoming commonplace in the clinic and changing outcomes for patients!

Soo Kng Teo is a scientist in the Geometrical Modelling department at the A*STAR Institute of High Performance Computing, Singapore.