16 MR US Fusion

Introduction

Thank you.

I'd like to thank Claire for that wonderful introduction.

The good news is she covered so much.

I think I can go pretty quickly through these slides.

Again, I have a travel disclosure.

Biopsy Planning vs. Surgical Planning

I think one of the first things to consider is how is biopsy planning different than surgical planning, for surgical planning?

As we've mentioned before, resolution of the capsule is of the utmost importance.

And therefore we generally encourage consideration for the use of an indirect coil for biopsy planning.

The most important thing is localization and suspicion grading.

And generally we can defer a coil and the full pelvis images.

Why Consider Biopsy Guidance

As Claire nicely delineated, why would we want to consider biopsy guidance?

Very briefly, 30% false negative systematic biopsy rate.

MR can identify cancers in over half of men with prior negative biopsies.

Direct MR guided biopsy provides imaging confirmation, but is not universally available.

In the face of a rising PSA negative biopsy, the second look is often warranted.

Also for follow up and active surveillance.

And where I can pinpoint the most suspicious area in the prostate and can be done either with direct in bore confirmation or with image fusion ultrasound guided biopsy.

And again, the technique here is optimized for detection rather than staging.

When to Consider Fusion Biopsy

When would we want to consider fusion biopsy?

MRI has not been proven yet to exclude significant cancer.

More significant disease is however found in targets systematics do find some significant disease missed by targets, both in our series and in the series from Dr. Poosh.

Most conventional ultrasound guided biopsies are systematic, not based on imaging findings.

And MR guided biopsies can target suspicious areas, but time on the magnet is rather costly.

So we fuse the MR data with ultrasound, and suspicious findings identified on multiparametric MRI can be co localized using ultrasound and biopsied.

That way this can expand image guided biopsies to patients where MR guided biopsies are unavailable.

And it means that we as radiologists may end up partnering with the urologists who do these biopsies.

Direct MR Guided Biopsy vs. Ultrasound Fusion Targeting

So when would you consider direct MR guided biopsy versus ultrasound fusion targeting?

Comparing the two, MR gives you direct imaging confirmation of targeting, but generally, image fusion is accurate to within three millimeters.

Direct inboard targeting can target all locations with equal ease, whereas the apex and far anterior lesions suffer from misregistration.

It takes about half an hour for setup, plus about 15 minutes for target for most inboard biopsies, whereas you can often do 12 systematic biopsies plus targets within 30 minutes using ultrasound.

However, we generally just do the targets when doing inboard biopsy, whereas it's fairly straightforward to add systematics to ultrasound, although you can do systematics with both systems.

With inboard targeting, we often use sedation, whereas with ultrasound fusion, we can perform a nerve block and use local anesthesia.

It requires dedicated hardware and software for both of these, and it's different hardware and software.

So you may want to only invest in one solution in terms of cognitive or mental fusion.

Again, Claire very nicely covered this.

Both systems provide improved targeting over systematic biopsies, but many of these targets can be seen with ultrasound.

In considering cognitive fusion, it's obviously a lot cheaper, but it does rely on understanding both MR and ultrasound features.

The initial investigations use this kind of approach, but there is a really elegant phantom study done by Mura a couple years ago where he found that for those targets that were invisible on ultrasound, almost a quarter or missed by more than three millimeters.

Available Systems

What are the systems that are currently available?

There's a very nice review last year by my colleague Leonard Marks, electric, sorry, external magnetic fuel tracking allows a like a Nintendo we style freehand manipulation, but you have to be careful for electromagnetic interference.

The mechanical arm can initially be cumbersome, but the arm itself stabilizes the probe and reduces motion error.

Image registration requires no dedicated hardware, but it provides only retrospective targeting.

So if you look at the comparison of the different types, you need new hardware for the articulated arm and for the EM tracking, you may only need new software for image registration.

You get real time target tracking with the mechanical articulated arm and with electromagnetic tracking.

However, with image registration, it's a step and shoot phenomenon where each time you move the ultrasound probe, you have to re-register the data set to determine whether or not you accurately targeting the lesion.

Patient motion requires a re-registration for both the articulated arm and electromagnetic tracking, whereas first off, for image registration, this is automatically compensated.

The mechanical arm can partially restrict motion, and as I mentioned, electromagnetic tracking is susceptible to electromagnetic interference.

However, with the image registration, one has to be careful about very steep angles where the image registration will fail.

The setup for the mechanical articulated arm requires attaching the arm, and from electromagnetic tracking, you have to register the tracker, for the software registration, this is basically automatic, and you may need to manually contour or re-contour the prostate.

And I'll show you an example of that for both the mechanical articulated arm and electromagnetic tracking.

But again, for the image registration, it's done automatically and you just have to confirm it.

Example of Prostate and Target Segmentation

Let's look at an example of prostate and target segmentation.

This can be performed either by the radiologist or if the lesion is identified by a technologist.

So you want to outline the prostate itself on one or more slices, and then you want to outline the target on one or more slices.

If you look step by step, the first step is prostate segmentation.

This may require more than one plane in order to accurately segment the prostate.

And so we usually segment in at least two planes.

And then once it generates a contour, you can see that you may need to adjust the contour slightly to refine it.

You then want to outline your region of interest.

So you can either outline the region of interest on every slice, or if the software supports it, outline on the middle, top, and bottom slices with interpolation.

For each region of interest, you want to identify it usually with a suspicion level.

This will be an example from a different platform.

The targets are provided in a three-dimensional presentation to the urologist, generally in a sort of a blob or a VR format.

You can see that the prostate is the brown blob, and the targets are the little blue blobs.

And those yellow lines through them are where the targeted biopsies were taken after the fact.

Here's an example before the targets are taken.

One of the advantages of using image fusion targeted biopsy is you can actually overlay a systematic biopsy template.

And that compensates for a lot of the problems where often the systematic biopsies are grouped or much closer together than the urologist thinks.

This records location of both systematic and targeted biopsies.

And so if a systematic biopsy were to be positive, even in the absence of a MRI abnormality, the urologist can go back and re-biopsy that area if the patient is undergoing active surveillance.

MRI Ultrasound Fusion Targeting Example

In terms of MRI, ultrasound fusion targeting, here's one example that illustrates the value.

This is a man where the PSA went from 2.5 to four, and there were systematic biopsies that showed less than 1% of one core Gleason three plus three.

This gentleman was given the option of active surveillance, and we were starting up a program where minimum active surveillance would have a biopsy to determine whether or not there was any suspicious area.

The repeat biopsy was done using MR Guidance.

Here's an example of a T two 80 image.

This T two contour was actually performed prior to our utilization of cad.

We saw this oval area uniform low signal.

Unfortunately, my region of interest obscures the margins, but they were circumscribed margins.

This was in the transition zone.

You've got what may be a lenticular lesion, which would now correspond to a PI-RADS four.

The ADC map showed corresponding restricted diffusion, and there was abnormal enhancement with washout characteristics.

This would be a type three curve, again, for a PI-RADS four or five lesion on T two, the DWI and the DCE do not affect the overall assessment category.

Based on this, we have a T two oval.

This would be transition zone mass that we give a suspicion category of four.

The ADC is focally restricted, but not greater than 1.5 centimeters.

So we would categorize it as four out of five.

The enhancement is positive, but again, we give this an overall assessment of four out of five.

This shows us, again, the lesion on the T two weight images, and this is how it's presented to the urologist.

It's flipped because it's confusing for them if we have the patient's right on their left side.

Here you see the ultrasound image with the prostate contour.

We can see that the delineation of the prostate is appropriate.

There's no misregistration, and here's the region of interest with the biopsy tracking through it.

Here's an image of the biopsy while it's being fired.

There is one other region of interest I didn't present to you.

Here's how it looks on the 3D images using parallax or two orthogonal planes in order to identify the lesions.

Both of the targeted cores were positive for low grade disease, however, both were high volume, so more than five millimeters or 80% of the core.

Because this was high volume, low grade disease, the patient opted for prostatectomy.

It turned out that on prostatectomy it was large volume, but organ confined, and low grade disease.

This was concordant.

The interesting thing is that it was such an obvious lesion on MRI, we generally do not often find low grade lesions on MRI.

Another Quick Example

One other quick example, this is a patient where in 2010, he had a multiparametric MRI, and you can see there's very slight T two asymmetry here.

It's nonfocal.

This would be given a T two score of two.

The ADC was really barely perceptible.

There was abnormal enhancement.

But this is basically a one or a two on the ADC map.

Biopsies at this time were negative.

A year later, we can see that this lesion is basically declared itself.

Now there's an obvious T two abnormality.

The ADC is focally restricted.

Again, the corresponding high b value DWI was bright.

This would be a four out of five assessment category on DWI.

And because this is a peripheral lesion, we use the DWI for the overall assessment.

Again, the perfusion was positive.

We see that multiple parameters are now highly abnormal and the targeted biopsy.

In this case, a patient also underwent systematic biopsies, but only the targeted biopsy was positive.

Processing a Case Example

If I can indulge you, I'd like to show you how we would process one of the examples.

As I pull the case up, I've already set my preferences to show a 3D MIP projection that's useful for localization, the T two ADC DWI.

And this will be the DCE map.

The first thing that I would want to do is make certain that the prostate contours are appropriate.

I can show that on my MIP as well as on the raw T two.

Now here you can see this is my pre contrast T one.

I review the pre contrast T one to evaluate for significant hemorrhage, and not seeing any, I'll apply either a pharmacokinetic map or a subtraction map.

We have one pharmacokinetic map here.

This shows you wash in and wash out characteristics.

But the other thing I can do is to use a simple subtraction map.

If I show the subtraction map, you see everything disappears 'cause we're subtracting zero from zero.

Here's an abnormality.

As I move the time cursor, I can identify the area of early wash in.

I'm gonna put the color overlay back on just to show that these two are concordant.

Okay, so I've identified something in the transition zone, and you can see that it's a oval fairly uniform low signal mass with irregular margins.

There's also obviously restricted diffusion and abnormal perfusion, but this being a transition zone lesion, we use the T two characteristics as the primary determinant.

In this case, I will outline the lesion, and I'll make this into a 3D region of interest.

I'll choose the top here and the bottom, maybe here.

And now it's a 3D region of interest.

I can identify the location on my maximum intensity projected image.

When giving this the location characteristics, I would choose the left anterior transition zone at the apex.

In fact, we can edit the ides info on a lot of these.

Here we have this is towards the apex.

I would choose this location.

This you can see is using the more conventional pi rads version one.

But as soon as the PI rads version two is published, we will likely see our industry collaborators update the automatic scoring.

I can evaluate the ADC value, although again, the quantitative value is not used in assessment.

I can also look at the DCE, which is obviously focally positive.

I can look at the washout curves, which are abnormal, but again, it does not contribute to the overall suspicion score.

This is an example of how you would process a case and then provide it for your refer the urologist to go ahead and biopsy.

One Other Example

I think I have a little more time, so I'll show you one other example.

The software will automatically pull up the hanging protocol.

I'll choose the prostate location.

Here we see another anterior lesion.

We very commonly see these in the anterior prostate because these are outside of the 1.7 centimeter standardized biopsy zone.

I'll go ahead and turn on my color overlay here.

You can see that this corresponds to focally abnormal perfusion.

It's a black and white lesion as yellow barrens might call it.

Let's see how big it is.

Again, it's a lenticular lesion with irregular margins in the really the anterior fibromuscular stroma.

But as we've mentioned earlier, this contains no glandular elements.

And so usually tumors in the anterior fibromuscular stroma arise from the transition zone.

We can see that it's more than 1.5 centimeters.

What I'll do this time is I will move, I'll copy the region of interest to the 3D image so we can see where the top and the bottom are.

In fact, I can propagate it to the other series.

If I choose my region of interest here on the DCE map, it will show me the curve characteristics.

But we already know this is positive.

Again, we already also know that it corresponds to the same area on DWI.

But because this is a suspicious lesion more than 1.5 centimeters, this would correspond to a PI RADS five score on T two.

This being the primary determinant would give us the overall score.

But the other thing I'd like to show you is if I simply choose the middle, bottom, and top slices, you can see that it shows me where they are on the 3D image.

When I'm done editing, it will fill it in.

And this provides a 3D target for the urologist.

Now you can see here we've got simple interpretation.

One of the things that we're looking for in the upcoming iterations of this software is a more smoothed algorithm for this interpolation.

Again, although this provides you as the radiologist all of the information you need to generate a report, including the prostate volume, which comes from the gland segmentation.

Here you see the prostate volume is 60 milliliters or 60 grams.

You also provide the image fusion targeted biopsy system with all the information needed for the fusion biopsy.

Thank you very much.