1

Jacob Wudtke
Clinical Internship III
CSI Treatment Planning
Having a background as a radiation therapist, my first exposure to treating patients with

CSI was with proton therapy which utilized 2 to 3 posterior static beams with 1 to 2 match lines.

From a therapist perspective, the setup process was simple, but there was a lot of potential for

mistreatment if the given shift parameters were not abided by during the setup/treatment process.

When I transitioned to UW-Health, to avoid the need for match lines or feathering CSI

treatments, CSI and TBI patients are planned on the Tomotherapy machine. Tomotherapy has

the capability of treating one continuous field if it is less than 130cm in length. Most CSI cases

are less than 130cm, however TBI cases are not, and in those situations, we treat headfirst and

feetfirst with a feathered junction. I have planned several TBI patients so far and will address

Tomotherapy feathering later.

The Tomotherapy machine at UW-Health only has 6 MV photon treatment capabilities,

which is what I would utilize if I was treating with VMAT on a Truebeam as well. My reasoning

behind utilizing 6 MV vs 10 MV is because there is target volume superficial with both the

spinal and cranial PTVs, and 10 MV gives rise to more superficial dose sparing, which we do not

want in this situation. Given that 6 MV is the treatment energy of choice, it is important to next

address the functionality of Tomotherapy and if not properly accounted for could lead to

variance in treatment daily.

Tomotherapy delivers radiation in a continuous helical fashion without the ability to split

beams into separate sections like a Truebeam. To combat this, structures can be created to block

entrance to areas of uncertainty or that are excessively sensitive to radiation like the lenses. In
 2

CSI cases, areas of uncertainty include shoulder positioning because they could have a different

superior/inferior position daily as well as arm position. If we created a plan that had no penalty

to the optimizer for entrance through the shoulders and arms, there is a good chance they will not

be positioned perfectly which will affect the dose distribution and potentially undercover the

target. The solution to fixing the shoulder issue is creating shoulder blocks which are shown in

figure 1 and 2 and blocking entrance through them to take shoulder uncertainty out of the

picture. The blocks I created extend superior past the shoulders 2 cm and inferior to where they

meet the arm blocks that were created.

Figure 1: Axial Plane Shoulder Blocks

 3

Figure 2: Coronal View of Shoulder Blocks and Arm Blocks

Now that entrance dose to the shoulders is taken care of, the next issue to address are the

arms. The arms are not reproducible unless the patient is positioned within a mold that has good

indexing. For cases such as CSI, it is more beneficial to keep the patient’s arms separated from

their body much like the patient in the case study because it allows for more room for beam

entrance without worrying about entering through the arms. If I solely created a contour of the
 4

arms and blocked entrance with no margins, it is impossible the patient will have their arms

perfectly positioned for daily treatment which would lead to potential dosimetric error. To

combat this issue, a 2cm margin is placed on the arms and entrance is blocked from that contour

which is shown in both axial and coronal view in figures 3 and 4.

Figure 3: Arm Blocks Axial View

 5

Figure 4: Coronal View Arm Blocks Abutting Shoulder Blocks

Figure 5: Protect Settings

 6

The final aspect to decide was pitch, time, and field width. Since Tomotherapy has

capabilities of 1cm, 2.5cm and 5cm jaws, there are always options when setting beams for

optimization. If one chooses a 1cm jaw the treatment time would take significantly longer than

the 5cm jaw. Since there are not many structures that have irregular shapes, to make time as

short as possible, I utilized a 5cm jaw which ended with around a 20-minute beam on time. Pitch

factor is what can cause a plan to look like it has a lot of threading. This factor is important to

address as a person plans because if we start with a high pitch factor (.430) the treatment can go

quicker, but at the expense of uniformity in the plan. The pitch factor I found to be most efficient

for time as well as plan quality was .315. Below are my beam specifics.

Figure 6: Beam Parameters

With the field being extensively long, there are many organs at risk such as cranial

structures including the lenses, eyes, optic nerves, etc. Glandular organs such as submandibular

and parotid are important to address. Thoracic and abdominal organs such as lungs, heart, liver

and especially the kidneys are important to address as well to avoid excessive dose as much as

possible. As addressed above, the lenses were spared by blocking entrance dose into them. The

kidneys were the next most difficult metric to meet while keeping uniformity due to their

proximity to the spine.

My plan was normalized to two separate volumes which are the PTVCranial and

PTVspine. Since this is one continuous volume, I did not need to utilize calc points, but if

junctions were needed, calc points would be beneficial to use. I set the prescription to try and
 7

envelope 95% of both PTV volumes to 100% of the prescription during the planning process.

This is how the prescription is usually set at UW-Health, and I figured since I have the

opportunity I would try and make it as clinically practical as possible. My final plan met all ideal

metrics and obtained 99.88% coverage of 95% prescription for the cranial PTV and 98.02%

coverage of 95% of prescription for the spinal PTV.

When evaluating the final plan, I was happy with the overall dose distribution as well as

dose uniformity. Figure 7 illustrates the dose distribution of the plan in coronal and sagittal view.

There is noticeable sharp dose falloff and the abdominal organs at risk are well spared when

referencing the sagittal view. To highlight the dose falloff around organs at risk, I have also

added figures in the axial view to show the fall off around the parotid glands, eyes, as well as

kidneys.
 8

Figure 7: Dose Distribution in Sagittal and Coronal View

 9

Figure 8: Dose Distribution Surrounding Parotid Glands

Figure 9: Dose Distribution Surrounding Eyes/Lenses

 10

Regarding hot spots, there were a few hotspots in the vicinity of the kidneys. The overall hotspot

is 39.5 which is acceptable in a case such as this, especially since it lies within the PTV. The con

of this location is that it is abutting the cauda equina, but the benefit is the cauda equina is more

radio resistant than the spinal cord, so this is a preferential location compared to it being in the

spinal cord. The hotspot region is illustrated in figure 10.

Figure 10: Plan Hotspot

With the prescription dose being 36 Gy and the max dose to the optic nerves also being

36 Gy, I needed to cool off that area to meet objectives. The optimization structure I utilized to

treat the cranial region cropped out the optic nerves which is shown in Figure 11. If I just

completely avoid the nerves, dose could drop excessively low and the PTV located in that region
 11

would be severely underdosed. To combat this, I created structures I labeled “Exclusion Zone”

(Figure 12), which are used to still push dose to that area with a min dose of 34 Gy and max of

35.8 Gy. This way the nerves are meeting the objective and the PTV is still getting an acceptable

dose. Figure 13 shows the dose distribution around the optic nerves as well as the cold spot.

Finally, Figure 14 shows the final DVH of the plan which is labeled accordingly. Regarding the

Proknow objectives, due to the methods described above, I was able to meet all objectives which

are outlined in Figure 15.

 12

Figure 11: Optic Nerves and Optimization Structure

Figure 12: Exclusion Zone

 13

Figure 13: Dose Distribution/Cold Spot Surrounding Optic Nerves

 14

Figure 14: DVH

 15

Figure 15: Proknow Certificate

Reflection

Of all clinical cases, CSI treatment planning can pose to be the most challenging. All the

organs at risk that surround the target make it difficult to provide a plan that is both impactful to

the patient but also safe. At UW-Health we are working on a workflow for Truebeam CSI

treatment planning as Tomotherapy is the standard. I have been working with physics to

troubleshoot and identify the best ways to junction fields as well as identify ideal beam

arrangements. I have learned that accounting for junctions can be more easily attained through

VMAT planning compared to 3D, as the optimizer can work well in smoothing out the cold/hot

spots.
 16

As mentioned before, I will note how we ordinarily junction treatments on a

Tomotherapy machine that have a field that is too long for one continuous field. We treat one

plan headfirst and the other feetfirst. The junction area between the two can give rise to varying

dose levels, so our only rule is that the junction is not located at the lungs or kidneys. Figure 16

shows how we draw out our junctions to match. It illustrates regular intervals of dose falloff, so

starting at the top section we go to 90% Rx in the optimizer and drop 10% as we move each level

inferior. For the opposite plan we essentially flip the percentages which gives a summation for

each individual junction of 100%. Although there are till small hot/cold spots, this has proven to

work well, and if I get a CSI plan in the future that is too large, I will use this technique.

Figure 16: Junction for CSI/TBI

 17

Overall, this lab was great because it required a different way of thinking. It made us

think outside of the box and apply a summation of multiple different techniques we have learned

into one. I am interested to read how other students planning approach turned out. I plan to

utilize these skills once I make the transition to dosimetry.