Transcript
Male Speaker: It's nice to be here to confirm or validate frameless base image guided radiosurgery and what Dr. Hamilton said. I'm sure there are some overlaps, some phantom studies we did a slightly different way. But recent advances in non-invasive immobilization devices in combination with image guidance, and 6-D table-top corrections enables us to treat benign or malignant lesions with high precision. The image guidance techniques are based on either stereoscopic or cone beam CT imaging devices. The patient is moved to the treatment position utilizing a 6-D robotic patient couch system. Other techniques based on surgical implantation of fiducial markers in the lesions were also clinically introduced and utilized. [Inaudible 00:01:21] Okay.
Male Speaker 2: Who stole the pointer?
Male Speaker 1: Thank you. The Novalis ExacTrac Version 5.2 6-D image guided frameless system enables us to image the patient at any couch position using Frameless Radiosurgery Positioning Array. Through stereoscopic image guidance, we can treat most, if not all, SRS procedures with frameless IGRS system. The aim is to dispense the invasiveness of the stereotactic frame fixation to the skull without losing the inherent accuracy of the frame-based stereotactic approach.
The purpose of this study is to evaluate, in phantom studies, the accuracy of the Novalis ExacTrac Version 5.2 6-D image guided frameless system. Based on our findings, we would establish a quality assurance program for the Novalis IGRS system. Furthermore, we report our initial experience with 15 patients who underwent frameless IGRS of trigeminal neuralgia.
The advantages of frameless radiosurgery as the previous speaker talked about it, first of all, the new BrainLab frameless IGRS system enables us to image the patient at any couch angle and correct for possible intra-fraction movements. This will eliminate the micro-adjustments that we had to do based on lasers and target positioning box. Basically, the other advantage is non-invasive, no pain, or no blood. No need for anesthetics and sterilization. It improves the patient's workflow, flexibility in treatment planning or treatment dates. And it also enables us to fractionate the treatments.
We use an in-house phantom which is made of Lucite with Kodak EDR2 radiographic film sandwiched between the two parts of the phantom. We made a mask for it. There was a 1.5 millimeter diameter spherical BB marker taped on the film. We will see here. There are two of them.
CT scan of the phantom was performed using 1.25 millimeter slice thickness in axial mode, and 512 by 512 matrix and the BrainLab head and neck localizer. The marker was defined as a target in the BrainScan 5.31 software. The same protocol that we use in our institution for treatment of trigeminal neuralgia which namely is 7 non-coplanar circular arcs with a four-millimeter cone is used to treat this marker using the stereoscopic image guidance.
The phantom was imaged at each couch angle and the images were fused with the DRRs using the autofusion option of the ExacTrac software. And ExacTrac x-ray correction was applied until the phantom was within an acceptance criterion of half a millimeter and half a degrees. The EDR2 film was developed and scanned using a Vidar Dosimetry-Pro scanner and RIT Technology software. The displacement of the centroid of the optical density on the film and the position of the marker, this was where the marker was positioned, was calculated using the RIT software. The same study was repeated for the second target.
Furthermore, retrospective study of 15 patients who underwent frameless image-guided radiosurgery of the trigeminal neuralgia since October of 2007 was performed. In those patients, we have done 186 stereoscopic imaging, which on average was like 12 per patient. And thir 6-D shifts were reviewed and reported here.
These are the seven different positions of stereoscopic imaging and trigeminal neuralgia patients, which basically you can see the nerve here. This red area is basically painted for the lower jaw to...since it's not the same position every day, so it excludes from the autofusion. And these are the other three positions. Most of the numbers that we get for shifts are if it's more than half a millimeter or half a degree is being corrected.
There are two shifts that we do. There are two sets of imaging we take. One is set for correction which basically when you position the patient to the isocenter, you have to correct that. And basically, that's the first position which is a bigger number. Well, we basically do verification. And if it's below one millimeter or half, it depends on...we usually use one millimeter or half one degree, then we start treatment. Most of the time if you have...like in this case, we have done 13 different imaging, and like in this case, we verified it here again and shifted at this point, and we verified it and we treated it.
Geometric accuracy of 0.48 ± 0.55 millimeter was achieved on two different targets treated in the phantom with a four-millimeter cone. The geometric accuracy of the target depends significantly on the landmarks used for other fusion in the ExacTrac software for these phantoms. And the average 3-D movement of the isocenter in the 15 patients post imaging and autofusion of the DRRs and the stereoscopic images was 0.65 ± 0.38. Most of the movement is in the longitudinal direction.
Now as we discussed in our last meeting, basically, the factors that affects overall precision of frame-based SRS system is your MRI, your CT, your fusion in the brain scan, then your lasers, the gantry isocentricity, your couch isocentricity, your target positioning box, and also finally, the patient movement. And this is based on this reference. Basically, the most inaccuracy that you get in this case is basically coming from MR, CT fusion to identify the nerve and overall errors that you get is 0.87 ± 0.41.
Now, if you would have done basically a couch micro-adjustment, then you would exclude that. You would bring it down to 0.63. Now when you are talking about frameless system, basically, you are still having the MRI, the CTA, the autofusion, but you also have to worry about the infrared markers or infrared detection systems, the x-ray and autofusion have the ExacTrac system as well as the gantry isocentric centricity and patient motion.
We have evaluated those new parameters which are here and it's very similar to the previous speaker's numbers which basically is 0.22 for autofusion and 6-D robotics is 0.2. Now, if you include all these parameters for image guidance, your total RMS frameless IGRS, the error you get 0.64 ± 0.39, which is still less than one millimeter. Now our phantom study showed that our error is 0.4 ± 0.55, which is, again, very similar.
So this is a comparison of frame-based as well as the frameless system. We are talking about 0.87 ± 0.41. And here we have 0.64 ± 0.39. If you adjust a couch with your microadjustments, you get 0.63, very similar. If you don't adjust it, obviously, image guidance system has a better accuracy.
One thing that we have to discuss also is patient movements, intra-fraction or inter-fraction. Basically, this is a patient that we have imaged. We have done imaging and we see that the laser is hitting different area or different parts of the... Basically, you have to do correction due to the patient motion here. There are three types of masks that is in use, but basically this is a standard perforated basic mask which is being used for whole-brain radiation. This is an IMRT-reinforced thermoplastic mask which is used for head and neck IMRT. And this is a BrainLab which has a mouthpiece and reinforced. Their price differences are very significant. This is probably about $200 and this is about $20. But in our experience, especially for trigeminal neuralgia, you wanna use this mask. It's reinforced. It's pretty steady. In our experience, it's been very consistent.
Well, based on our findings, we would like to basically recommend or we do this following QA which basically the weakest link in the entire procedure is MRI or MR fusion which basically you have to make sure you don't have distortion. And your physicist in diagnostic radiology usually checks the resolution distortion using a phantom. And we keep it below about 0.3 millimeter. And obviously, it significantly depends on your matrix size.
The X, Y, Z gradient also has to be checked. The CT image resolution phantom should be less than 0.3. Gantry isocentricity using Winston-Lutz test is less than...it should be less than 0.25. And the infrared camera calibration as well as the 6-D robotic couch and ExacTrac autofusion, we do it daily and it's typically less than 0.2 millimeter. And if you basically do these and do this test, your overall error should be less than 0.7 millimeter.
This is our MRI QA of bimonthly. You can see that the Y axis gradient is not consistent and it had to be corrected as well as the overall field homogeneity. So in conclusion, accuracy of the target localization achieved with the BrainLab frameless technique approaches that attainable with invasive stereotactic frames. However, since ExacTrack 5.2 autofusion software heavily relies on the bony landmarks, special care must be made with a frameless system by taking at least one set of stereoscopic images for correction and one verification x-ray image set per bed position to ensure accurate treatment of the trigeminal nerve. Thank you.
Male Speaker 2: Who stole the pointer?
Male Speaker 1: Thank you. The Novalis ExacTrac Version 5.2 6-D image guided frameless system enables us to image the patient at any couch position using Frameless Radiosurgery Positioning Array. Through stereoscopic image guidance, we can treat most, if not all, SRS procedures with frameless IGRS system. The aim is to dispense the invasiveness of the stereotactic frame fixation to the skull without losing the inherent accuracy of the frame-based stereotactic approach.
The purpose of this study is to evaluate, in phantom studies, the accuracy of the Novalis ExacTrac Version 5.2 6-D image guided frameless system. Based on our findings, we would establish a quality assurance program for the Novalis IGRS system. Furthermore, we report our initial experience with 15 patients who underwent frameless IGRS of trigeminal neuralgia.
The advantages of frameless radiosurgery as the previous speaker talked about it, first of all, the new BrainLab frameless IGRS system enables us to image the patient at any couch angle and correct for possible intra-fraction movements. This will eliminate the micro-adjustments that we had to do based on lasers and target positioning box. Basically, the other advantage is non-invasive, no pain, or no blood. No need for anesthetics and sterilization. It improves the patient's workflow, flexibility in treatment planning or treatment dates. And it also enables us to fractionate the treatments.
We use an in-house phantom which is made of Lucite with Kodak EDR2 radiographic film sandwiched between the two parts of the phantom. We made a mask for it. There was a 1.5 millimeter diameter spherical BB marker taped on the film. We will see here. There are two of them.
CT scan of the phantom was performed using 1.25 millimeter slice thickness in axial mode, and 512 by 512 matrix and the BrainLab head and neck localizer. The marker was defined as a target in the BrainScan 5.31 software. The same protocol that we use in our institution for treatment of trigeminal neuralgia which namely is 7 non-coplanar circular arcs with a four-millimeter cone is used to treat this marker using the stereoscopic image guidance.
The phantom was imaged at each couch angle and the images were fused with the DRRs using the autofusion option of the ExacTrac software. And ExacTrac x-ray correction was applied until the phantom was within an acceptance criterion of half a millimeter and half a degrees. The EDR2 film was developed and scanned using a Vidar Dosimetry-Pro scanner and RIT Technology software. The displacement of the centroid of the optical density on the film and the position of the marker, this was where the marker was positioned, was calculated using the RIT software. The same study was repeated for the second target.
Furthermore, retrospective study of 15 patients who underwent frameless image-guided radiosurgery of the trigeminal neuralgia since October of 2007 was performed. In those patients, we have done 186 stereoscopic imaging, which on average was like 12 per patient. And thir 6-D shifts were reviewed and reported here.
These are the seven different positions of stereoscopic imaging and trigeminal neuralgia patients, which basically you can see the nerve here. This red area is basically painted for the lower jaw to...since it's not the same position every day, so it excludes from the autofusion. And these are the other three positions. Most of the numbers that we get for shifts are if it's more than half a millimeter or half a degree is being corrected.
There are two shifts that we do. There are two sets of imaging we take. One is set for correction which basically when you position the patient to the isocenter, you have to correct that. And basically, that's the first position which is a bigger number. Well, we basically do verification. And if it's below one millimeter or half, it depends on...we usually use one millimeter or half one degree, then we start treatment. Most of the time if you have...like in this case, we have done 13 different imaging, and like in this case, we verified it here again and shifted at this point, and we verified it and we treated it.
Geometric accuracy of 0.48 ± 0.55 millimeter was achieved on two different targets treated in the phantom with a four-millimeter cone. The geometric accuracy of the target depends significantly on the landmarks used for other fusion in the ExacTrac software for these phantoms. And the average 3-D movement of the isocenter in the 15 patients post imaging and autofusion of the DRRs and the stereoscopic images was 0.65 ± 0.38. Most of the movement is in the longitudinal direction.
Now as we discussed in our last meeting, basically, the factors that affects overall precision of frame-based SRS system is your MRI, your CT, your fusion in the brain scan, then your lasers, the gantry isocentricity, your couch isocentricity, your target positioning box, and also finally, the patient movement. And this is based on this reference. Basically, the most inaccuracy that you get in this case is basically coming from MR, CT fusion to identify the nerve and overall errors that you get is 0.87 ± 0.41.
Now, if you would have done basically a couch micro-adjustment, then you would exclude that. You would bring it down to 0.63. Now when you are talking about frameless system, basically, you are still having the MRI, the CTA, the autofusion, but you also have to worry about the infrared markers or infrared detection systems, the x-ray and autofusion have the ExacTrac system as well as the gantry isocentric centricity and patient motion.
We have evaluated those new parameters which are here and it's very similar to the previous speaker's numbers which basically is 0.22 for autofusion and 6-D robotics is 0.2. Now, if you include all these parameters for image guidance, your total RMS frameless IGRS, the error you get 0.64 ± 0.39, which is still less than one millimeter. Now our phantom study showed that our error is 0.4 ± 0.55, which is, again, very similar.
So this is a comparison of frame-based as well as the frameless system. We are talking about 0.87 ± 0.41. And here we have 0.64 ± 0.39. If you adjust a couch with your microadjustments, you get 0.63, very similar. If you don't adjust it, obviously, image guidance system has a better accuracy.
One thing that we have to discuss also is patient movements, intra-fraction or inter-fraction. Basically, this is a patient that we have imaged. We have done imaging and we see that the laser is hitting different area or different parts of the... Basically, you have to do correction due to the patient motion here. There are three types of masks that is in use, but basically this is a standard perforated basic mask which is being used for whole-brain radiation. This is an IMRT-reinforced thermoplastic mask which is used for head and neck IMRT. And this is a BrainLab which has a mouthpiece and reinforced. Their price differences are very significant. This is probably about $200 and this is about $20. But in our experience, especially for trigeminal neuralgia, you wanna use this mask. It's reinforced. It's pretty steady. In our experience, it's been very consistent.
Well, based on our findings, we would like to basically recommend or we do this following QA which basically the weakest link in the entire procedure is MRI or MR fusion which basically you have to make sure you don't have distortion. And your physicist in diagnostic radiology usually checks the resolution distortion using a phantom. And we keep it below about 0.3 millimeter. And obviously, it significantly depends on your matrix size.
The X, Y, Z gradient also has to be checked. The CT image resolution phantom should be less than 0.3. Gantry isocentricity using Winston-Lutz test is less than...it should be less than 0.25. And the infrared camera calibration as well as the 6-D robotic couch and ExacTrac autofusion, we do it daily and it's typically less than 0.2 millimeter. And if you basically do these and do this test, your overall error should be less than 0.7 millimeter.
This is our MRI QA of bimonthly. You can see that the Y axis gradient is not consistent and it had to be corrected as well as the overall field homogeneity. So in conclusion, accuracy of the target localization achieved with the BrainLab frameless technique approaches that attainable with invasive stereotactic frames. However, since ExacTrack 5.2 autofusion software heavily relies on the bony landmarks, special care must be made with a frameless system by taking at least one set of stereoscopic images for correction and one verification x-ray image set per bed position to ensure accurate treatment of the trigeminal nerve. Thank you.