VMAT and IMRT Flashcards
what is fielf in field?
forward-planned IMRT
-leaves modulate the beam intensity; leaves are static while gantry rotates and while beam is delivered
-dose not delivered while gantry moves
what is step and shoot
like fif but inverse planned
IMRT
-dose not delivered while gantry moves
what is sliding window
inverse planned IMRT where the leaves are moving as dose is being delivered
-dose not delivered while gantry moves
The first leaf is always moving. To decrease the intensity at a location, the 2nd leaf will move with it (to get a downward slope). Hence the complexity of the first leaf determines the max leaf speed for that slice. To get an upward slope, the first leaf has to move faster.
what is varied in VMAT to achieve the desired dose distribution?
dose rate first, then gantry speed
fluence based optimization vs aperture based optimization
-IMRT may be either or
-VMAT is aperture based
-if fluence based, a separate leaf sequencing step is necessary
what are dynamic conformal arcs?
MLCs change shape with moving target
difference between IMRT and VMAT dose distribution
• Compared to IMRT, a VMAT distribution will generally be similar for the dose levels > 50%. However, the VMAT plan will generally have lower doses distributed more uniformly throughout the normal tissue due to having beams coming in from all directions
If you can’t use acurods, what should you use for a lesion in lung?
TG 101 doesn’t allow for pencil beam to be used in inhomgeneous regions
-if can’t have acuros at least use AAA
-uses point dose-spread kernels, which perform density scaling for the distance between the interaction point and the calculation point, thereby assuming that electrons travel in a straight line along this direction.
explain volumetric normalization
-used by VMAT- want certain volume to get certain % of dose- can adjust final normalization to make plan hotter/colder by 2 %
-not like 3DCRT where we normalize to a point (but still plan such that X volume receives Y dose)
Does VMAT require a CT?
yes because beams are coming in from all directions
In contrast, with IMRT, can potentially use BEV images
• Describe the process of progressive resolution optimization used in VMAT as described by Otto. How has this been implemented in the Eclipse planning system?
o Continuous gantry and MLC leaf motion is modelled by a coarse sampling of discrete gantry angles and MLC apertures. VMAT involves progressively increasing the gantry and MLC position sampling resolution as the optimization progresses. Later on in the optimization, smaller changes in MLC position and MU weight per gantry angle are allowable due to the mechanical and efficiency constraints which limit variation in dose rate and MLC aperture shape from one gantry angle to the next. Each iteration involves randomly choosing a MLC leaf or MU weight to modify. If the change is allowable given the constraints, then the dose distribution and cost function are calculated. If the cost is reduced, then the change is accepted. Transition to the next level is determined by how much the cost function has plateaued as a function of time.
o In Eclipse, there are four levels of optimization, with more gantry angles at higher levels. The number of control points remains fixed throughout optimization. Within PO (which has replaced PRO), dose is quickly calculated (for the purpose of calculating the cost function) using MRDC (multiple resolution dose calculation). In MRDC, variable gantry angle resolution and multi resolution scatter computation is used.
MRDC is within PO/PRO- like AAA but resolution is variable
ou have been assigned the project of implementing VMAT for head and neck treatment in a facility that does not yet use this technique. How would go about it? What resources (human, equipment, etc.) would you need?
o Would need to do a series of end-to-end tests to make sure entire process from simulation to delivery is running properly.
o Testing synchronization of gantry position, leaf position and dose rate.
o MLC leaf positioning accuracy, reproducibility.
o Dosimetry tests (e.g., flatness, symmetry and output as a function of gantry speed & dose rate)
o Since it is no longer step and shoot, must expand Q&A routine to account for the fact that the beam is on while the gantry rotates, and while the MLC (and possibly also the collimator, depending on implementation) is moving.
o Verify mechanical constraints on the machine (VMAT optimizer requires this information).
o Interruption/resumption tests
o Leaf leakage characterization (interleaf, through leaves, through leaf ends)
o Absolute dose measurements with a calibrated ion chamber to make sure TPS calculates dose properly to a point.
o Compare plan objectives obtainable with VMAT to plan objectives achievable with old method (e.g., IMRT) to ensure that acceptable plan quality is achievable; comparison with VMAT treatments from the literature.
what makes VMAT unique from previous implementations of therapies involving continuous gantry motion while beam is on?
VMAT involves progressively increasing the gantry and MLC position sampling as the optimization progresses
why do you need to sometimes rewind to an earlier level with VMAT?
less samples, more parameter flexibility, less constraint sometimes required to achieve bigger changes
Approaches to IMRT
-optimize fluence, then figure out MLC leaf sequence needed
- or else, start with BEVs with apertures and equal dose rates. At each optimization step, change MU weight or MLC leaf position
Max MLC leaf speed, collimator jaw speed, and gantry speed
2.5 cm/s for MLC/jaws, and 6 degrees/s
why is it good for jaws to track with MLC leaves?
minimizes effect of inter-leaf, midleaf leakage
does optimization prefer to change dose rate or gantry speed?
dose rate
gantry has a lot of inertia-lass resort
definition of cost function
sum of terms that are quadratic dose differences multiplied by a priority value
when does progression to next optimization level occur in PO/PRO?
when cost function has plateaud
-at higher levels, plateus moer quickly because due to mechanical contraints smaller changes are possible
how are MU weights of new MLC positions related to their neighbours?
-positions are linearly interpolated from adjacent samples
-MU weights of new samples are a function of MU of neighbours
when does progression to next optimization level occur per Otto?
doesn’t consider levels
-considers exponential more iterations between sample additions when there are fewer samples
-doesn’t consider plateauing of cost functions
Does optimization per Otto use simulated annealing?
No
who used simulated annealing
IMRT- optimizaions that don’t have progressive sampling (i.e. used fixed number of samples)
how densely do you need to sample to ensure < 5 % of volume has dose errors > 3 %
1 degree for gantry rotation, and 0.5 cm for MLC leaf motions is required
advantages of aperture based optimization
-reduce modulation
-consider contraints at time of optimization; don’t need extra step
why is more rigid immobilization required in IMRT?
higher dose gradients- setup accuracy more vital compared to ex. POP
is error in contouring more or less rlevant in IMRT vs 3DCRT?
more, because the contours directly control the result
in 3DCRT, contours were only spatial guides
how are PTV and PRV margins determined?
based on population statistics of comparisons between the planning CT (obtained from CT-sim) and daily pre-treatment verification CTs obtained over the course of treatment
how do parotids change over course of treatment?
translate ~1 mm/week medially and undergo ~20% volume loss over Tx course
difference between forward and inverse planning
o Forward planning:
Set beam orientations, weights, modifiers
Forward dose calculation
Check if PTV and OAR dose volume constraints are satisfied – return to step 1 if not
o Inverse planning:
Set beam orientations: choose treatment ports (IMRT) or range of gantry angles (VMAT)
Specify desired dose volume constraints for PTV(s), OAR(s) – specify optimization objectives for contoured structures, including tuning structures/pseudocontours (not representative of a real anatomical structure; used to guide the optimizer to the desired solution).
Inverse plan using optimization techniques e.g., VMAT
Forward dose calculation
Check if PTV and OAR dose volume constraints are satisfied – return to step 1 or 2 if not
what can cost functions include?
target: actual dose- prescription dose, minimum and maximum dose (latter are step functions)
OAR: DVH constraints, max constraints (step functions)
determinist vs stochastic optimization methods
• Gradient-based optimization methods are deterministic; can get stuck in local minimum. Simulated annealing optimization is stochastic (doesn’t necessarily yield the same answer every time); can escape from local minimum
-does this by once in awhile accepting a solution that increases the cost function
max dose volume
0.03 cc or 0.1 cc
-need VOI to be at least one voxel
what happens to dose outside of contoured OARs?
has to be looked at after optimization
VMAT considers it with NTO (normal tissue optimization)
benefits for VMAT vs IMRT
-faster (2 min vs 10 min)
less chance of patient movement and intra-fraction physiological motion
downside of VMAT
more time consuming to plan and QA
why do we sometimes us 90 colli?
bilateral disease
using 90, some leaves can go in between the targets and improve conformity
think of treating vertebrae that wrap around the cord- can get half of bone with one arc and cover cord with MLC, and other half with another arc (cover cord with MLC)
why can’t colli be at 0?
doesn’t smear out the inter-leaf leakage
what is volumetric normalization?
-in VMAT, 95% of PTV sould get 100% of prescription dose
comparison of VMAT and IMRT dose distrubtion
generaly similar for isodose > 50 %
VMAT plan will generally have lower doses distributed more uniformly throughout the normal tissue due to having beams coming in from all directions
difference between VMAT and previous implementations of therapies involving continuous gantry motion while the beam is on
VMAT involves progressively increasing the gantry and MLC position sampling as the optimization progresses
Issue with using full gantry range is that we are limited by MLC motion constraints between discrete gantry positions. Hpwever, if we don’t sample enough of the range we won’t get a good plan or a good indication of the dose distrubtion (discrete approximation not a good approximation of continuous case)
o To tackle this tradeoff, VMAT uses progressive sampling of gantry and MLC positions.
o Sometimes need to rewind to earlier level (with less samples, more parameter flexibility, less constraint) to achieve more drastic changes. Bigger adjustments are made in leaf sequencing during the initial phases of optimization.
classic approach to IMRT/VMAT and alternative approach
classic: optimize fluence maps, then figure out MLC leaf sequence needed to achieve this fluence
alternative: start with a series of BEV aperture shapes that are conformed to target minus OAR (e.g., prostate excluding rectum). This is what is used by VMAT. Initially, dose rates for all segments are equal. At each optimization step, either MU weight of a particular segment, or a MLC leaf position (these are the two optimization parameters that are adjusted) is changed.
types of constraints for VMAT optimization
o Mechanical constraints ensure that the aperture shapes and MU weights are physically achievable: e.g., overlapping of opposing leaves and nega
o Efficiency constraints: MLC leaf motion and MU weight constraints that ensure continuous and timely treatment delivery – want to avoid gantry deceleration; want to rotate gantry at maximum possible speed as much as possible. Need to consider maximum travel speed of leaves (2.5 cm/s) and maximum gantry rotation (6 deg/s), max allowable dose rate [these limits are values for Truebeam].
Also note the collimator jaws maximum speed is 2.5 cm/s (same max speed as MLC leaves). In our clinic, our collimator jaws track with the MLC leaves to ensure the smallest possible aperture, to minimize the amount of interleaf, intraleaf and leaf end leakage (due to leaf pairs not being able to be perfectly closed) that reaches the patient.
Otto seems to ignore max allowable change in dose rate between control points. This assumes that dose rate can change fast enough that we don’t need to worry about it as a constraint.
tive MU weights are not physically possible.
