3DVH does not require the CT of the patient for accurate calculations but does utilize CT data for display purposes.

3DVH maps the errors from phantom to patient using a process called Diff-Morph, which is a key part of the patented 3DVH process.  This algorithm has been researched and published in many journals.  In white box tests (where the outcome is known), 3DVH showed remarkable accuracy – the DVH results from 3DVH almost perfectly overlie the known results.

Sun Nuclear carefully considered various approaches for determining dose in a patient volume before developing 3DVH.  The ideal approach was to use measured data and the original treatment plan data and not re-create a new forward calculating algorithm.

Forward calculating algorithms for QA have the inherent problem in that one is never sure which algorithm to believe if there is a difference – the QA algorithm or the TPS algorithm.  Many TPS and forward calculation QA providers also share intellectual property suppliers – therefore the same error could exist in both the TPS and QA system providing a misleading agreement (i.e. a false negative).

Finally, TPS algorithms have benefitted from many years of research and continuous improvement – it is unlikely that a forward calculation engine for a QA product would prove more accurate or robust than a highly tested and vetted TPS engine.  3DVH uses the TPS data as a starting point and then perturbs the treatment plan based solely on measured data, and compares measured 3D dose to the expected treatment plan dose, without adding another algorithm as an extra variable and possible source of error.  The most elegant solution is often the best.

3DVH is very accurate — there are several publications^{4, 5, 6, 7, 8} that have analyzed the accuracy of 3DVH, all of which have had excellent results. To prove that the algorithm works there are several approaches.

The approach that is most similar to a patient measurement would be to take measurements with film and numerous chambers in a heterogeneous phantom, and verify that the 3DVH results were equivalent to the measured results.

Another approach is to already know the correct answer and test if 3DVH reproduces this answer (i.e. White Box Testing). This is performed by introducing an error in a treatment plan file, but delivering the treatment correctly. If 3DVH correctly reconstructs the correct treatment plan using the correct delivered plan dose with the erroneous treatment plan, then we can state with confidence that 3DVH is accurately reconstructing dose based on measured data. This approach was published⁴ by the University of Wisconsin.

4. "Moving from gamma passing rates to patient DVH-based QA metrics in pre-treatment dose QA"
   Med. Phys. 38, 5477 (2011)
5. "3D DVH-based metric analysis versus per-beam planar analysis in IMRT pre-treatment verification"
   Med. Phys. 39, 5040 (2012)
6. "Evaluation of the accuracy of 3DVH software estimates of dose to virtual ion chamber and film in composite IMRT QA" Med. Phys. 39, 81 (2012)
7. "On The Accuracy Of The Planned Dose Perturbation Algorithm"
   Sun Nuclear white paper (2010)
8. "PDP Accuracy in the Presence of Significant Tissue Density Variation"
   Sun Nuclear white paper (2010)

No. Unlike secondary calculation methods, 3DVH does not require commissioning or machine data. Simply select your Linac, MLC and energy combination to load a pre-configured model.

More information is always a good thing, though it can be intimidating.  Clinical Practice improvement is the goal we all strive for, and 3DVH is a tool to achieve that goal.  Many clinics have used 3DVH and found and corrected hidden systemic errors that were affecting all patients, which is an incredible benefit.  As with IGRT, not knowing was/is easier, but patients can greatly benefit from a correctional shift prior to treatment.  Patients can likewise greatly benefit from 3DVH determining what is actually being delivered prior to treatment. By finding systemic errors, or simply errors within a specific plan BEFORE a patient is treated, 3DVH represents a tremendous step forward in radiation therapy.

3DVH does not offer that functionality, but there should already be several safeguards to catch CT to Electron Density curve errors.  Monthly CT QA (per TG-66) and Annual QA should catch any CT to Electron Density errors.  Because these errors are easily caught using a phantom with various known densities, this is not a problem that would require a full forward calculation model to address.  

More severe heterogeneity errors can be imagined if the TPS algorithm is simply not robust enough to accurately calculate around heterogeneous tissue interfaces.  This is a general problem (not patient specific) and should be discovered when commissioning a new TPS algorithm.  A physicist could easily detect these problems by using any heterogeneous phantom and performing an end-to-end test where multiple points are measured around heterogeneities.  This is a TPS commissioning task, and should not need to be repeated for every patient.

The internal dose grid size for ArcCHECK based 3DVH (AC-PDP) is < 2mm, and for MapCHECK 2 based 3DVH (MC-PDP) is 1mm. 3DVH reconstructs the internal dose grid at a very precise, high-resolution so that the software does not have to interpolate results and introduce uncertainty.

The high resolution dose grid allows comparison of the 3DVH and TPS results without having to interpolate or scale the 3DVH results.  The TPS data is left in its original state (pixel size) as received from the TPS.

DVH bin size refers to the size of each dose bin used in accumulating the DVH statistics (i.e. binning dose points into the histograms).  3DVH assigns dose bins using the maximum dose of each plan in order to use the highest number of bins for a given plan.  3DVH uses the following bin sizes:

• For global max < 1 Gy, bin size = 0.00001 Gy;
• 1 < global max < 10 Gy, bin size = 0.001 Gy;
• 10 Gy < global max < 100 Gy, bin size = 0.01 Gy
• Global max > 100 Gy, bin size = 0.1 Gy.

From this, the number of bins is at minimum 1000 and maximum 10,000, assuring high resolution of DVH bin width and QuickStat computation.

No. The magnitude of the dose error is used in addition to the depth and patient surface characteristics to adjust the dose across the patient volume.  These corrections are derived from the basics of the Compton effect.  Detailed discussion of this process can be found in a publication⁹ from the University of Wisconsin.

9. "Moving from gamma passing rates to patient DVH-based QA metrics in pre-treatment dose QA"
   Med. Phys. 38, 5477 (2011)

This is a clinical question that clinician teams will have to discuss in order to determine the correct answer for their clinic, much like physicians/physicists had to do when IGRT was implemented. The following is an example of the kinds of clinical implementation protocols that could be followed:

The physician could determine thresholds of DVH differences for PTV, Cord, and other OARs.  If the threshold (e.g. 3% difference in PTV at 95%Rx; 200cGy difference in the Cord Max dose) was exceeded in the physicist’s review of the 3DVH results, the physician could be asked to review the DVH and make the clinical decision whether to proceed.  At that point, either the plan could be revisited (or scaled if all organs were cold/hot), with the results re-run through 3DVH.  Throughout this process physics could be responsible for overall practice improvement if systematic errors were discovered.

These are all clinical decisions that must be agreed upon within the clinic; this suggestion is given only as an example, not as a guide.

All common combinations are supported, and no commissioning is required for 3DVH, simply select the model for the specific setup used. New combinations can be added when requested. Available PDP models include:

• Elekta 4MV (80 Leaf MLC)
• Elekta 6MV (80 Leaf MLC)
• Elekta 6MV (Beam Modulator MLC)
• Elekta 8MV (80 Leaf MLC)
• Elekta 10MV (80 Leaf MLC)
• Elekta 10MV (Beam Modulator MLC)
• Elekta 15MV (80 Leaf MLC)
• Elekta 18MV (80 Leaf MLC)
• Elekta 25MV (80 Leaf MLC)
• Siemens 6MV (160 Leaf MLC)
• Siemens 6MV (58/82 Leaf MLC)
• Siemens 10MV (58/82 Leaf MLC)
• Siemens 15MV (58/82 Leaf MLC)
• Siemens 18MV (160 Leaf MLC)
• Siemens 18MV (58/82 Leaf MLC)
• Varian 4MV (120 Leaf MLC)
• Varian 6MV (120/120HD/80 Leaf MLC)
• Varian 6MV TrueBeam FFF (120/120HD/80 Leaf MLC)
• Varian 10MV (120/120HD/80 Leaf MLC)
• Varian 15MV (120/120HD/80 Leaf MLC)
• Varian 18MV (120/120HD/80 Leaf MLC)
• Varian 23MV (120/120HD/80 Leaf MLC)

The 3DVH perturbation calculation takes approximately 2-3 minutes for RapidArc plans with 2 Arcs.  For IMRT plans the calculation is usually less than 1-2 minutes.  What physicists gain from these few extra minutes is very valuable to their patient - a clinically relevant IMRT QA result.

Once adopted, 3DVH will save time because the metrics provided by 3DVH are more intuitive, sensitive, and specific than passing rate metrics.  Passing rate metrics, whether they pass or fail, tell the physicist/doctor very little about the clinical fitness of a specific treatment plan, which can leave a dosimetry team guessing what needs to be altered when plans do fail passing rate criteria.  The continual practice improvement that comes with adopting 3DVH will assist the entire treatment team in efficiently producing excellent treatment plans.

ArcCHECK was specifically designed to expose a large and consistent number of SunPoint Diode Detectors to the beam regardless of the gantry angle.  ArcCHECK not only has a surface that is always normal to the beam, it also allows for two measurement depths – entrance and exit dose.  This is the ideal arrangement for VMAT QA and is required for 3DVH with VMAT.

A 2D array must be positioned in one of three ways: sitting on the couch with the VMAT beam delivered 360 degrees around it, attached to the Linac head, or in a rotisserie so that the 2D array is always perpendicular to the beam.  All arrangements pose problems. Couch top loses much of the data density because the array is not always normal to the beam.  This means that at angles such as 90 or 270 degrees, there are very few diodes collecting data and because there’s no entrance and exit dose (as in the AC), 3DVH doesn’t even know what Gantry angle the data is coming from.  The second and third scenario also suffers from the Gantry angle problem – with the array attached to the Linac (or using the EPID), 3DVH has no way of knowing were the data is coming from geometrically.    Some manufacturers use a mechanical inclinometer, but this is user and surface dependent, and would not produce results that were trustworthy enough to be used in the 3DVH calculations. Additionally, a secondary mechanical inclinometer introduces more setup time.

The Virtual Inclinometer independently calculates gantry angle to +/- 0.5 degrees, without additional cables or ancillary hardware. The angle is calculated using the divergence observed between entrance and exit dose passing through the ArcCHECK cylinder¹⁰. The Virtual Inclinometer’s primary use is to provide the required gantry angle data necessary for the 3DVH reconstruction.

10. “Optimizing the accuracy of a helical diode array dosimeter: A comprehensive calibration methodology coupled with a novel virtual inclinometer”
    Med. Phys. 38, 5021 (2011)

The Virtual Inclinometer is an accurate and useful tool and is available for ancillary uses.  There are two primary uses for the Virtual Inclinometer functionality beyond 3DVH:

1. VMAT Monthly QA - the physicist can perform monthly VMAT QA using the same, complex VMAT plan each month.  Use the Virtual Inclinometer to confirm that the gantry vs. time at several points remain constant.  (e.g. The first month record the gantry angle at time = 20, 40, and 60 seconds.  The following months confirm that the gantry reading is remaining constant at those time intervals.)


2. Static Gantry Angle Monthly QA – in place of the standard Gantry angle test of using a level and recording the Gantry angle at 0, 90, 180, and 270, one can generate a treatment plan with four static open fields.  Using the Virtual Inclinometer, the actual gantry angle could be recorded versus the nominal gantry angle.

The MLC patterns come directly from the treatment plan.  When analyzing a 3DVH result that has unexpected errors, seeing the leaf patterns of the plan can sometimes offer a clue as to whether the plan was over-modulated or not.  If not over-modulated, the physicist can move on to other potential causes, such as modeling issues that would be more visible in the BEV tab. The main practical use of the 4D Workspace is for the physicist to quickly and easily review the dynamics of the plan they are analyzing.  The physicist can quickly check:

• Modulation patterns
• The number of monitor units per beam/plan
• The number of control points
• Review the gantry motion vs. MLC leaf motion.

Physicists can also review the ArcCHECK entrance/exit dose (in conjunction with the MLC patterns) as a function of time for qualitative analysis of delivery dynamics..

Control Point Analysis is an ArcCHECK function that helps the physicist determine where an error is originating from a gantry angle perspective.  VMAT QA taken as a whole arc is actually composite dose QA, and errors can blur into one another and be hidden.  When Control Point Analysis is used, VMAT QA becomes analogous to Beam by Beam IMRT QA, which may highlight beams/sub-arcs that are not being delivered accurately.  For example, if the Couch isn’t modeled correctly the posterior part of the plan will likely have errors.  Or if the MLC is drifting due to gravity, the lateral portions of the plan will have erroneous readings.  Control Point Analysis takes composite QA back to a Beam by Beam analysis so that the physicist can more easily pick up on sub-arc specific errors.

BEV is more useful for IMRT field analysis since the BEV for a rotating (VMAT) beam is really a composite of all gantry angles’ fluences. The 3DVH BEV tab will still show these composite BEV maps in order to keep the GUI uniform and consistent no matter what plan type.