Mark Kendrick

DOS 523 – Treatment Planning in Medical Dosimetry

Treatment Planning Paper

April 28, 2019

Planning radiation therapy treatments for lung tumors with or without inhomogeneity

correction results in differences in dose distribution to the target volume, and also to the

surrounding normal tissues1.

Standard isodose charts and depth dose tables were created in reference to radiation beam

performance in specific conditions, including homogeneous material such as water, or

soft tissue. When the beam encounters tissue with density variations, such as air, bone,

metal, or odd shaped combinations of tissues with various densities, it causes a change to

the isodose lines and change the dose distribution1.

There are two ways in which tissue inhomogeneity affects dose distribution. The first is

that it changes the attenuation of the primary beam the pattern of photon interactions

downstream. The second is the way the electron fluence changes as a result. The greater

the density of the inhomogeneity, the greater the attenuation and loss of primary beam

energy, but also increased secondary electron fluence. Conversely, if the inhomogeneity

is less dense, there is less attenuation to the primary beam, so it maintains higher energy

at greater depth while secondary electron fluence decreases1.

In this case, we are looking at lung tumors and the inhomogeneity involved in treating

them. As the beam traverses the lower density of the lung tissue, the beam maintains a

higher energy, or depth dose compared to the standard isodose charts, or what would be

expected in water or soft tissue. This results in higher doses to the lung than would be

expected, unless we corrected for the inhomogeneity with a factor for lung density.

A second issue affects dose distribution to the lung tumor, particularly if the mass is

situated such that the beam traverses lung tissue prior to contact with the mass. In this

case, the secondary electron fluence decreases and requires penetration of some depth,

generally greater than 1.5cm for 6 MV beam energy, before the normal electron fluence

would reach the expected level that it would have in water or soft tissue. In other words,

the electron scatter component of the dose would need to build up with depth upon

transitioning from lung tissue to tumor volume.

A third issue for dose coverage to a lung tumor is that, for a small lung tumor, to be

treated with a small field size, there would be an increased dose fall off near the edges of

the field due to scattered electrons escaping the lateral beam edges before depositing their

energy. This is because they would travel farther in the lower density lung tissue than in

water or soft tissue, allowing more of the scattered electrons to escape the beam laterally.

This decreases the dose due to the scattered electron component around the periphery of

the tumor from the beam’s eye view1.

My clinical site, Renown Institute for Cancer, uses Acuros XB to calculate dose for lung

tumors. This algorithm is very strong in its capability to account for dose in

inhomogeneous tissues. In my very limited experience planning, or observing lung

treatment planning at Renown, I have never seen the heterogeneity correction turned off.

The one lung treatment plan that I was involved in was an SBRT treatment, and

inhomogeneity correction was turned on. With the capabilities of Acuros in predicting

radiation interactions with matter, we rely on the dose calculations produced by Acuros.

We have a high level of trust in the capabilities of Acuros to present accurate predictions

of dose distributions involving inhomogeneity including within, adjacent, and

downstream from the inhomogeneity. For this reason, we leave it turned on. The

physicians, in turn, must accept the physical reality of some of the limitations in dose

delivery with inhomogeneous scenarios such as lung tumors.

The only times that we override the inhomogeneity values from the CT scan are either:

1. When there is high-density artifact present, such as a metallic artificial joint or

dental implant, causing the CT values to misrepresent the real tissue density, and

must be overridden with a homogeneous Hounsfield unit (HU).

2. When there is a contrast agent, or other radiopaque material present that will not

be present during treatment, and must be overridden with a homogeneous

Hounsfield unit. In these cases, the volume containing contrast is overridden and

assigned a HU value similar to an average of the surrounding tissue and material.

Within the field of radiation oncology, the debate continues, whether or not to use

heterogeneity correction2. Proponents of using the correction argue that more accurate

predictions of dose distribution resulting from inhomogeneity corrections are essential.

Those who have reservations about utilizing inhomogeneity corrections express concerns

due to the idea that treatment regimens have been established and refined under
conditions in which inhomogeneity corrections were not used. The concern is that

although inhomogeneity corrections may result in more accurate dose calculations, their

use may still result in dose deliveries that differ from those which were used to establish

the successful treatment regimens, and possibly result in less favorable patient outcomes.

Yet, radiation oncology is a field that is rightfully determined to operate with the greatest

possible accuracy, to the point of sub-millimeter, in some cases. More accurate methods

will prevail and become the norm as time passes, and older, less accurate methods will

fall to the wayside.

One clinical scenario in which severe inhomogeneity exists is when a patient has metal

inside of them as they undergo CT simulation, treatment planning and treatment delivery.

A study by Maerz et al.3 found that severe streaking artifacts caused by metallic objects

within patients impact CT data and inhibit accurate representation of shape and density of

the high density object(s) and the surrounding tissue. This study determined that the

resulting impact to calculation inaccuracy could be as significant as an increase or

decrease in dose calculation of 11% versus a corresponding film measurement.

Another study by Ziemann et al.4 found that inaccuracies resulting from streaking related

to high density objects resulted in calculation inaccuracy in the order of 8.4%. They also

found that streaking correction algorithms in some CT software packages could

significantly improved planning accuracy, and could reduce this number to a much more

reasonable 2.7% inaccuracy due to the same high density artifact (HDA).

In order to compare and evaluate the use of inhomogeneity correction in the treatment

planning environment, I will set up a test patient with a lung tumor as the target volume.
The inhomogeneity will involve the presence of lung tissue adjacent to target volume.

With my treatment plan designed, I will calculate the dose in the default setting – with

inhomogeneity correction turned on, and I will calculate again with inhomogeneity

correction turned off. This will reveal the differences between calculations, which

attempt to account for dose distribution with the inhomogeneity scenario involved with a

lung tumor, and calculations, which instead perceive the patient as a singular,

homogeneous tissue density, like a column of water.

According to Essers et al.5, lung treatments are more accurate when incorporating a

heterogeneity correction model. My clinical site tends to agree, so we use heterogeneity

correction, but I will perform a comparison to test and evaluate the differences to enhance

my understanding of the differences between lung plans generated with or without use of

the correction.

To set up a comparison between lung treatment plans with inhomogeneity correction

turned on, and off, I completed the following steps: I imported the patient’s DICOM data

and structures to Eclipse. In the contouring tab, I placed a carbon support structure under

the patient. In Eclipse, I created a new plan with a dose of 6000 cGy prescribed to the

PTV. I placed an AP beam with 6 MV energy. I created a new MLC and shaped it to the

PTV with a 1cm margin. I created a new opposed field from the PA, with 6 MV energy.

I created a new MLC for the PA field and shaped it to the PTV with a 1cm margin. I

noticed sloping isodose curves toward the patient’s superior and lateral sides, so I placed

10 degree dynamic wedges on both fields with heels toward the superior and lateral sides

to even out the isodose curves. This should level the isodose curves to produce a more
uniform dose distribution, and result in a reduced hot spot value after normalization. I

selected Acuros XB 13.5 for the calculation model and algorithm. I turned off the

heterogeneity correction. I calculated the plan. I observed the dose distribution, then,

adjusted beam weighting to achieve balanced coverage and minimum hot spot. I

normalized 100% dose to 95% of the PTV. I check for optimum beam weighting again

to minimize the hot spot. I copied/pasted the plan and turned on the heterogeneity

correction. I calculated the plan, and then adjusted the beam weighting to achieve

minimum hot spot. Now, I have two plans, with and without heterogeneity correction,

for comparison.

Looking at the dose distribution between the plan with inhomogeneity correction on, and

the plan with inhomogeneity correction off, there are some apparent differences in

appearance. First, the plan with inhomogeneity correction off has much more geometric

and symmetric appearance to the dose distribution. The curves of dose distribution in the

axial views are more smooth and rounded. The dose distributions on the coronal and

sagittal views are more smooth and representative of depth dose curves. The dose

distributions for the plan with inhomogeneity correction on are less smooth, have

variation and detail in the isodose lines, representative of a greater variation in

attenuation due to the accounting for the range of tissue densities that the beams traverse.
Figure 1: Inhomogeneity ON

By comparing two treatment plans for a lung tumor, with and without inhomogeneity

correction, I was able to enhance my understanding of the differences between the two

approaches as it applies to dose distribution to the target volume, as well as to the

surrounding normal tissue. By normalizing 100% of the dose to 95% of the target

volume, both plans achieved identical target coverage values. I was able to identify

differences in hot spot and surrounding normal tissues. I was also able to identify

differences in the appearances of dose distribution between the two plans, with much

more variation and details apparent in the plan with inhomogeneity correction turned on,

and smoother, more symmetrical isodose lines, more closely representing depth dose

curves, in the plan with inhomogeneity correction turned off. These observations have

not changed my opinion that we produce more accurate treatment plans with

inhomogeneity correction activated.

 References

1. Khan FM, Gibbons JP. The Physics of Radiation Therapy. Philadelphia: Lippincott

Williams & Wilkins; 2014.

2. Papanikolaou N, Klein EE, Hendee WR. Heterogeneity corrections should be used in

treatment planning for lung cancer. Medical Physics. 2000;27(8):1702-1704.

doi:10.1118/1.1287645.

3. Maerz M, Koelbl O, Dobler B. Influence of metallic dental implants and metal

artefacts on dose calculation accuracy. Strahlentherapie Und Onkologie: Organ Der

Deutschen Rontgengesellschaft . [Et Al]. 2015;191(3):234-241. doi:10.1007/s00066-014-

0774-2.

4. Ziemann C, Stille M, Cremers F, Buzug TM, Rades D. Improvement of dose

calculation in radiation therapy due to metal artifact correction using the augmented

likelihood image reconstruction. Journal Of Applied Clinical Medical Physics.

2018;19(3):227-233. doi:10.1002/acm2.12325.

5. Essers M, Lanson JH, Leunens G, Schnabel T, Mijnheer BJ. The accuracy of CT-

based inhomogeneity corrections and in vivo dosimetry for the treatment of lung cancer.

Radiotherapy And Oncology: Journal Of The European Society For Therapeutic

Radiology And Oncology. 1995;37(3):199-208. https://search-ebscohost-

com.libweb.uwlax.edu/login.aspx?direct=true&AuthType=ip,uid&db=mnh&AN=874658

8&site=ehost-live&scope=site. Accessed April 25, 2019.