The “Best-in-Physics” poster session at the American Association of Physicists in Medicine (AAPM) Annual Meeting highlights the top five abstracts in the imaging, therapy and multi-disciplinary science tracks. In 2020, however, as with many events this year, things were different.
This year’s annual meeting, held jointly with the Canadian Organization of Medical Physicists (COMP), went virtual. Instead of crowds gathering around the posters, presenters shared their research via talks presented on the meeting’s online portal. Here is part one of my selection from this year’s top scoring studies at the 2020 Joint AAPM|COMP Virtual Meeting. Look out for a second report later this week.
MRI-Linac enables simultaneous MLC tracking of two moving targets
Paul Liu, from the ACRF Image X Institute at the University of Sydney, described the use of an MRI-Linac to simultaneously track the motion of two treatment targets. “There are many radiotherapy cases that involve simultaneous treatment of multiple targets,” he said, citing examples such as locally advanced prostate or lung tumours, and oligometastases.
The challenge here is that both targets can move independently of each other. Currently, this is addressed by using large treatment margins, but this then confers extra dose to healthy tissue. Liu explained that the improved imaging capabilities of an MRI-Linac can be used to localize multiple targets simultaneously.
Liu and collaborators used the Australian MRI-Linac, which has a 1T magnet and a 120-leaf multileaf collimator (MLC), to track two spherical targets irradiated with a 6 MV conformal field. They used a motion platform to test three sinusoidal motion traces, plus lung and prostate motion traces recorded from patients.
Multi-target tracking is achieved via an extra pre-treatment aperture segmentation step, Liu explained. An MLC tracking algorithm uses the target volumes, geometries and MLC positions from the treatment plan to divide the MLC aperture into two segments specific to each target.
During treatment, the team recorded 4 Hz cine MR images and used a template matching algorithm to calculate the motion of each target. When motion was seen, rather than shift the entire MLC aperture, the leaf positions of each segment were calculated independently and then recombined into a single MLC aperture encompassing the motion of both targets.
For sinusoidal motion, the team recorded a tracking latency between the targets and their corresponding apertures of 328 ms, comparable to single-target MLC tracking systems. Liu noted that two-thirds of this time is related to the MRI step, potentially enabling latency reduction by using faster imaging and reconstruction techniques
For lung motion traces, “multi-target tracking does a good job overall in tracking both targets simultaneously,” said Liu. The geometric uncertainty – defined as the RMS error between the target and aperture positions during treatment – was reduced from 5.5 mm without tracking to 2.7 mm. After correcting for latency, this error reduced further to 1.2 mm. For prostate traces, the RMS error was reduced from 4.2 mm with no tracking to 1.4 mm with multi-target tracking.
Putting these findings into clinical context, simulations showed that in both cases, large margins of up to 7 mm were needed to maintain target coverage without tracking. With multi-target tracking, 3 mm margins provided over 99% coverage.
“This is the first experimental demonstration of tracking two independent moving targets on an MRI-Linac,” Liu concluded. “We showed that this technology can help reduce margin sizes in cases of differential motion.”
Quantitative ultrasound evaluates plaque vulnerability
Carotid plaques within arteries are at risk of rupture, which can cause adverse health effects such as cognitive deficits or stroke. But assessing plaque vulnerability – how likely it is to break off, travel to the brain and cause hypoxia – is difficult, as the mechanisms of plaque rupture are unknown and each plaque is unique in size, shape and composition.
“Our research tries to improve the ways that we can noninvasively and inexpensively see the carotid plaque that can cause stroke or other harmful heath events,” explained Catherine Steffel from the University of Wisconsin-Madison. “We want to give clinicians an inside look at which plaques are vulnerable.”
Steffel and colleagues are investigating the use of quantitative ultrasound imaging to noninvasively identify a plaque prone to rupture via its scattering structures. Their goal was to define quantitative ultrasound parameters that can identify plaque composition and help assess its risk.
Steffel examined the effective scatterer diameter (ESD) – a parameter that describes the frequency dependence of acoustic backscatter and has been used previously to examine other tissues. She notes that her team’s previous studies on excised plaque showed that small ESD corresponded to calcified regions, while larger ESD was seen from lipid particles in plaque.
The researchers performed in vivo ultrasound imaging of carotid arteries in 52 patients scheduled for clinical removal of plaque. Just over half of the participants displayed symptoms of stroke or transient ischemic attack and they had a median stenosis (artery narrowing) of about 74%. After the plaque was surgically removed, it was scored by a surgeon and assessed by a pathologist. The team then compared the in vivo images with the surgical scores and histopathology assessment.
Analysis revealed weak correlations between ESD and surgical calcification rating, and between ESD and histopathology cholesterol/fibrinoid level. Stronger correlations were seen between ESD and histopathology hemosiderin score. Steffel explained that the hemosiderin scores are important for evaluating plaque composition as hemosiderin deposits are key markers of plaque vulnerability.
Best in physics: clinical prompt gamma measurements and FLASH dosimetry
“Our results for calcium and cholesterol/fibrinoid do not agree with previous literature, perhaps because our parameters were computed for the entire plaque rather than a region-of-interest,” said Steffel. “The hemosiderin findings intrigued us, as to our knowledge, this has not been studied previously.”
In particular, the team found that two parameters – the average ESD and standard deviation ESD – could differentiate between hemosiderin scores of 0 and 2. “Overall, this work demonstrates that ESD parameters may help identify plaque composition and microstructure, providing a future clinical tool for assessing plaque vulnerability with noninvasive, relatively inexpensive ultrasound imaging,” Steffel concluded.
