Before we start with motion management let's remind ourselves what is understood as contemporary conventional photon radiotherapy. Certainly a standard radiotherapy today is three-dimensional and conformal, using multileaf collimators whether for beam intensity modulation or beam shaping only. Also image guidance of treatment delivery should be considered standard today.

Radiotherapy process starts with patient CT imaging in treatment position. This provides a patient's three-dimensional model on which treatment plan is prepared. Treatment planning is actually beam simulation using sophisticated computer programs. The process output is a treatment plan, i.e. set of linac parameters to deliver optimal dose distribution, and reference images.

    Reference images are compared with live verification images acquired using dedicated X-ray imaging system on a treatment machine to make sure patient treatment position is as close as possible to the one on planning CT. Position verification is performed using either a pair of 2 orthogonal 2D images or 3D cone-beam CT (CBCT). 2D verification images are compared with digitally-reconstructed radiographs (DRRs) from planning CT, 3D cone-beam CT is compared with planning CT directly.

    Comparison process objective is bone anatomy, or metallic markers implanted in tumour vicinity, or, when visible, a tumour itself. In such case we speak about soft-tissue match

    Like with everything in life - plan is one thing and reality another. Even with image guidance, and also based on chosen objective, tumour position can differ more or less from planned position. This is also due to process (in)accuracy, but mainly due to difference of planning CT model and reality. We are living creatures and we change continually so naturally one image in time cannot be expected to represent whole anatomy accurately for long. This results in differences in dose delivered to target and organs-at-risk.

        Very important in radiotherapy is the concept of target volumes by ICRU - International Commission on Radiation Units & Measurements. This picture is from their No 62 report demonstrating

GTV - gross tumour volume - as a structure actually visible for given imaging modality

CTV - clinical target volume - GTV expanded by sub-clinical involvement of the disease, meaning 'what is not visible but likely to be there'

ITV - internal target volume - CTV expanded by margin corresponding to expected internal motion, and finally...

PTV - planning target volume - with another expansion accounting for setup uncertainty

This picture demonstrates general efforts for reduction of margins to enable either higher tumour dose or lower normal tissue toxicity.

Inter-fractional variation/motion...

       ...as mentioned before, due to changes in organs position, size or shape, tumour is generally  misplaced from planning position every fraction. Image-guidance is used to correct for this displacement assuming no change in dose distribution shape. Luckily, photon dose distributions are relatively robust so this aspect is usually negligible.

    Target can change not only its position....also it can rotate and/or deform. Also patient's shape can change for various reasons. Therefore, original plan based on a single 'snapshot' CT model does not necessarily guarantee correct dose delivery for all fractions.

    The key tool to address inter-fractional variations is IGRT - image-guided radiotherapy. It is important to distinguish between using IGRT for target daily position correction only, or using it also for treatment plan validity check. In general, the more difference we see between verification image and planning CT, the bigger differences between dose delivered and planned we can expect.

To deal with this problem there is a concept of adaptive radiotherapy. The simplest example of this approach is patient rescan and replan when significant deviation is detected.

For ultimate adaptive radiotherapy solution one needs to evaluate combined delivered dose distributions of two and more fractions. This is impossible without Deformable Image Registration...

    ...this is an example of two CTs for a lymph node near stomach. Due to difference in stomach content there is a significant difference in position and shape of both the target and critical organ. To combine effect of doses delivered in these two situations, deformable image registration is needed. In general, image registration is equivalent to finding a mathematic transformation mapping each voxel on one image to identical voxel (element of tissue) on another image. Rigid registration consists of translations and rotations only. Non-rigid registration allows for deformations too. This is a challenging scientific problem but there are commercial programs available today...

Intra-fractional variation/motion... 

    Respiration - is natural human process associated with relatively large internal motion mostly in chest and abdomen regions. The table shows review of respiration motion amplitudes in 3 anatomy directions based on published papers. Usually motion in cranio-caudal direction is larger compared to other directions and can be as large as few centimeters.

Also there is a nice demonstration of trajectories of number of tumours in lung with exact site indicated.

    However, respiratory is not the only source of intra-fractional motion. Heart beat is another regular physiological process but there are also rather random processes. For example, these can be associated with sudden move of bowel or rectum due to gas. This is typical for prostate treatments

    There are two principles used to address sudden unpredictable motion:

First, it is minimizing time between verification image with positional correction and actual beam delivery.

Alternatively, technology can be used to monitor target position nearly in real time. This is based on implanted markers in tumour vicinity. These can be either active, i.e. signal transmitting, used in commercial products such as Calypso or Raypilot, or passive, used for Cyberknife treatment for example.

Cyberknife beam direction and position is corrected based on frequent imaging of implanted markers using a pair of X-ray imagers. Cyberknife plan typically has over 100 beams and position update can be done for each beam if needed.

    On the other hand, respiration is rather regular periodic process. It causes variation of tumour and adjacent organs positions, their shapes and, in case of lung, also density. This all represents challenges for accurate treatment planning

    In order to handle respiratory motion, the key is to either monitor tumour position directly, which is very challenging, or indirectly using surrogates. These can be either internal - typically active or passive metal markers, or external. External surrogates monitor breathing phase and are based on various principles such as infrared markers placed on chest, imaging patient surface in 3D, pressure belts, or those based on measuring or controlling air volume in lung.

External surrogate signal correlation with position of either tumour itself or an internal surrogate may/should be verified ideally both before & during treatment

    These are examples of commercially available respiratory monitors. Two based on air volume control, 2 based on tracking surface using infrared markers, one based on surface topography and one pressure belt.

    At the London Clinic we use Varian RPM system with gaming goggles for patient feedback. Using goggles significantly helps patient to achieve desired and reproducible breath hold. Imaging and treating at breath hold is one of motion management strategies - to virtually stop respiratory motion for time needed

    Probably most important tool in addressing respiratory motion in radiotherapy today is four-dimensional CT (4DCT). More principles exist - the one demonstrate here is the axial/cine acquisition phase-binned 4DCT

- RPM signal is recorded over the whole examination period

- For each table position, projection data for breathing period+1s is acquired

- CT reconstructs an image every given interval, e.g. 0.5s, which for 5s period generates approximately 10 images

- Each peak-to-peak segment of RPM record is split in 10 phase intervals per cycle. Using  time vs. phase relationship given by RPM record the image reconstructed at time  corresponding to nearest phase is selected for each desired phase. This, done for each table position generates a single phase 3D CT series -> 10 single-phase CT series = 4DCT

    This is an example of thorax 4DCT.

    4DCT is not only a series of 3D CTs for various breathing phases which can be played as a movie to assess motion range. Useful CT series can be reconstructed from 4DCT. Maximum intensity projection (MIP) and average intensity projection (AIP) images are obtained as maxima or average voxel signal over all single-phase components of 4DCT. They are used to guide contouring or as a representative 'average' model for dose calculation.

Also single-phase series can be extracted or derived to represent mid-ventilation (midV) or mid-position (midP) as  situations best representing tumour position in time or in space.

    Impact of 4DCT is mainly in having information about respiratory motion range for treatment planning. Other two motion management strategies relate to dose delivery control. First one is GATING when beam is ON only when target is at desired position or phase, the other is TRACKING with treatment beam following target during the whole cycle

    The two major motivations for respiratory management are

          - to increase accuracy of treatment planning and dose delivery, and...

          - to reduce treatment margin or treated volume

With standard approach the position of tumour is a 'snapshot' - random somewhere in tumour trajectory during breathing. To cover 100% of the tumour we have to apply estimate of full motion range both directions.

If we have a 'slow' CT for lung or, better, 4DCT, with the same intent we can treat volume which corresponds to ITV - internal target volume.

To reduce treated volume significantly, there is gating or tracking. These are all approaches assuming patient's free-breathing during treatment.

Alternatively we can treat at breath-hold - either at inspiration or even deep inspiration, or at expiration which is considered more reproducible.

    If we accept that tumour is not covered by treatment isodose all the time during respiratory cycle and determine target volume and margins using rather statistical methods, comparison of treated volumes for various methods may look different. This concept however, should not mean compromised tumour dose or tumour control probability.

    Now let's have a look at the impact of respirtory motion on IGRT using a simple phantom with regular motion and 2mm ball bearing in the centre of the cube.

    This is an anterior and lateral DRR from a standard CT. The ball bearing image is about 5mm long - slightly more than reality and also more than as on three kV images acquired on a linac at 3 different 'breathing' phases

    DRRs reconstructed from the average intensity projection CT, obtained from 4DCT, provides information about ball bearing motion range in both directions - 10mm in anterior view and 16mm in lateral view, respectively.

    Also relevant is comparing standard fast CT with a standard slow cone-beam CT on a linac where acquisition takes 30 or 60 seconds based on acquisition mode. Difference between a 'snapshot' ball bearing on the first image set, and rather average nature of CBCT is clearly visible.

    Average CT derived from 4DCT provides a CBCT equivalent image from motion perspective. Unfortunately, both images are also blurred because of motion.

    ...finally, selecting a single representative phase from 4DCT is, from image quality perspective, equivalent to 'snapshot' standard fast CT.

These are motion aspects of currently available imaging modalities having an effect on quality and accuracy of IGRT

Review of respiratory motion management strategies

    A choice of motion management strategy depends on clinical objective and also on technology available.

If patient needs to be treated at free-breathing, treated volume doesn't need to be necessarily reduced to spare critical organs, and CT model representativeness is not an issue - then standard, i.e. no motion management, is the choice

If any of these aspects are relevant however, patient may profit from employing some motion management strategy including 4DCT  and ITV, treating at mid-ventilation or mid-position, gating at free-breathing or breath-hold, and tumour tracking as technologically most challenging option.

    Let's have a closer look at key aspects of a standard - no motion management approach:

    - standard 3D CT is acquired and used for treatment planning

    - margins are conventional, i.e. mostly standard WITHOUT assessing patient individual respiratory motion

    - for IGRT, the most accurate in principle is daily CBCT with tumour or soft tissue match

    - alternatively, a 2D/2D marker match can be used

    - bone match, either 2D/2D or 3D is only adequate if there is confidence about margins covering potential motion

    - treatment delivery is standard at free-breathing

    For ITV concept the most important difference relative to standard is target including tumour motion derived from 4DCT...

    For CTmidV or CTmidP approaches, the planning model represents most likely or mid position of the tumour and margins are derived using statistical methods but include patient's individual characteristics...

    Gating at breath-hold means that planning CT, pre-treatment imaging and also treatment delivery are performed at desired breath hold...

    Gating at free-breathing is similar in principle but planning CT is acquired either prospectively at desired phase, or extracted from the full 4DCT...

    Tracking, as implemented in Cyberknife, is different not only in principles but also by intrinsic validation of correlation between internal surrogate - fiducials - and external surface surrogate. The correlation model is established prior to treatment and is verified frequently during treatment delivery. It is used to predict tumour position in time, and direct treatment beams accordingly...


    ...the slide shows respiratory motion magnitude versus margins applied for motion management strategies presented. American association AAPM recommends motion management applied from motion range of 5mm and above....this newer publication suggests using 4DCT from 8mm on, and using gating from 13mm motion magnitude on...


       The key aspects of motion management from treatment planning perspective:

        - understanding & handling 4DCT including artefacts

        - image registration of multiple series with motion assessment objective in addition to target definition objective

        - understanding motion aspects of IGRT including application of relevant reference structures to guide matching

        -  margins specific to various motion management strategies

        -  understanding limitations of using external surrogates

Used and recommended literature...