My experience with using the module

[mrtig]

Hi,

I promised to write about my experience a long time ago, now I finally have some time to do it.
I'll try to describe the process of surgery planning.

The general workflow

1. Receiving the CT scans of the mandibula and the lower extremities in the DICOM format.
2. Meeting with the surgeon(s) to discuss where to make the cuts and how many segment
3. Segmentation of the mandibula and fibula
4. Performing the virtual surgery using the Bone Reconstruction Planner
5. Asking for confirmation from the surgeon(s)
6. 3D printing the reconstructed mandibula
7. Modelling the mandibula and fibula guide(s) in Blender
8. 3D printing the surgical guides

Each workflow has been a bit different. The number and order of steps can vary.
At first I was only 3D printing mandibulas so the surgeons could use them to pre-bend the metal plates. Sometimes the mandibula is the first thing I print before doing any planning because it will be used as a backup in case there are some issues with the virtual plan and it's also nice for the surgeon to mark the cut lines on the mandibula.

Preparation of the data in Slicer

1. Crop the CT images so they would only include the regions of interest
2. Using the segment editor's threshold tool segment both the mandibula and one of the fibulas.
3. Using the island tool and scissors remove he parts that are not part of the fibula and mandibula.
4. Transform the fibula segmentation so it's next to the mandibula.

Using the Bone Reconstruction Planner

The progress of the module has been so fast. I last used in in the end of February- beginning of March. As I was almost always in a hurry when doing the planning when I didn't have time to figure out how to do something I used brute force methods to get the job done.
So I am not really sure how useful it would be to describe how I used it.

The plans

Case 1

Here only one segment of the mandibula had to be removed.

The guides were enterily designed in Blender. I only used the surfaces generated by the Bone Reconstruction Planner (BRS) as the basis.

The guides only have one guide surface.

I tried to use the features of the bone so the guides could be accurately positioned.

The shortcoming is that that the mandibula guides are not connected. This increases the chance of error. In another case I made guides that are connected.

Case 2

The removed piece of the mandibula was replaced by 3 fibula segments.

The guides are still in 2 parts and with one guide surface

I got the feedback that the right guide was not so good in terms of accurate postioning

The fibula guide

Case 3

These are the most advanced guides. I used the miter boxes on the fibula guide and a connecting tube between the mandibula guides. The mandibula guide also has an angled cut in the ramus area so the coronoid process, which is difficult to remove and doesn't need to be removed, stays in place.

The connecting tube should increase the accuracy of positioning of the guides.

These are the features I used for positioning.

Case 4

This case is a bit more andvanced also and I have to at least edit the guides in Blender.
The end result would be something like this:

So only a small portion of the ramus would be left and it wouldn't be an end to end joint, but the side of the fibula fragment would be joined with the end of the ramus.

I was not able to make the guides for this case because of lack of time but I will try to finish this case just as an exercise.

I made the fibula guide base by copying the fibula to a new segmentation, then used the margin tool to grow it by 3 mm and finally substracting the original fibula from the guide base. Only thing left was to trim the guide base to size.

Case 5

This time I was able to create the fibula guide almost completely in Slicer. I only made some small modifications in Blender.
The problem was that the boolean didn't work perfectly and parts of the miter boxes intruded through the guide base.
I also tried to create the mandibula cut guide in Slicer, but gave up on it when I tried to create a tube that joins the cutting guides. I don't know if it's because I don't know how to manipulate fiducials in slicers or it's a feature - I was not able to place the fiducials where I wanted them to be. My expectation was that the fiducials would move only in a plane that is parallel to the view. That's not the case. When I moved the fiducial up, down, left or right, it also moved closer or farther away in respect to my current view. I tried for 15 minutes and then made the tube in Blender.

[mauigna06]

Hi @mrtig. Thank you for writing about your experience using the BoneReconstructionPlanner.

I just would like to clarify that you used BoneReconstructionPlanner to make the fibulaSurgicalGuide in case 3.

Now it is possible to make mandibleSurgicalGuides also. See #36 for more information. If you want you can give feedback about how sawBoxes are positioned with the automatic positioning algorithm.

[mauigna06]

Referring to case 4.

I was not able to make the guides for this case because of lack of time

So you only used BoneReconstructionPlanner for virtual surgery planning and to make the fibula surgical guide? Was the fibula surgical guide used to perform the surgery?

This case is a bit more andvanced also and I have to at least edit the guides in Blender

What did you have to edit and why?

[cmfsx]

@mrtig is the 'green' segment from the fibula segment or is it the original mandible?

[mauigna06]

BRP could be used to plan how the fibulaPieces reconstruct the neomandible (like is shown in the picture, no need to transform them by hand). I think the green bone is part of the mandible made with the segment editor to simulate the resection on that side of the mandible because the first cut plane (to the left of the dark grey bone piece) cuts too much.

Using the plan of the neomandible, you should have been able to make the fibula surgical guide with BRP.

I made the fibula guide base by copying the fibula to a new segmentation, then used the margin tool to grow it by 3 mm and finally substracting the original fibula from the guide base. Only thing left was to trim the guide base to size.

But I see this and it is not clear to me if you did or did not make (and print) the fibula surgical guide for this case

[mrtig]

@cmfsx - the green is the original mandible.

@mauigna06 No guides were printed.

[cmfsx]

@mrtig Thank you for your kind reply.

@mauigna06 I was thinking of the mandibular angel reconstruction. I think the way we create the angle (with ramus from fibula) is very good.

But in the case the @mrtig highlighted (original ramus and fibula ends at the angel ) and if we want to create two segments on top (like in the photo) we have few problems. The first case can be done by using two planning setups then combine. But in the other case, I was thinking if we can put a 'critical angle' beyond which the VSP take it as to put the other segment on top of the other ?

So when the critical angle is reached then the segment is cut straight ( from the origin of the plane leave around 5 mm ?) and the next segment is put on top at the end.

[mauigna06]

In the future using a new type of cut planes could be done to achieve this:

With some more work double barrel planning and surgical guides could be possible

Before that you could also give priority to dental implants planning and surgical guides

[mauigna06]

Hi Martin. Thank you for keep logging your uses of BoneReconstructionPlanner. Could you post some pictures of this last case? Maybe the reconstruction and the guides?

This time I was able to create the fibula guide almost completely in Slicer. I only made some small modifications in Blender.
The problem was that the boolean didn't work perfectly and parts of the miter boxes intruded through the guide base.

Could you attach some pictures that show what the problem was? Please try creating a new issue about this.

I also tried to create the mandibula cut guide in Slicer, but gave up on it when I tried to create a tube that joins the cutting guides. I don't know if it's because I don't know how to manipulate fiducials in slicers or it's a feature - I was not able to place the fiducials where I wanted them to be. My expectation was that the fiducials would move only in a plane that is parallel to the view. That's not the case. When I moved the fiducial up, down, left or right, it also moved closer or farther away in respect to my current view. I tried for 15 minutes and then made the tube in Blender.

Yes. The fiducials are a little tricky to position on empty space. You have to move the camera a lot to make little adjustments to achieve the position you want.

[mrtig]

Hi,

Sorry for not posting pictures. I just didn't have time. I posted what I could so it wouldn't get lost. I will take a look at the case again when I have time and I'll post some screenshots.

[lassoan]

My expectation was that the fiducials would move only in a plane that is parallel to the view. That's not the case. When I moved the fiducial up, down, left or right, it also moved closer or farther away in respect to my current view. I tried for 15 minutes and then made the tube in Blender.

You can choose between moving the points in the view plane or snap to visible surface in Markups module: Display / Advanced / 3D display -> unconstrained (moves in view plane) or snap to visible surface.

If you choose "snap to visible surface" and you create a model around the point and the model's Selectable property is True, then indeed, it becomes impossible to place the curve, because it will just run into the camera.

[mrtig]

Case 5

This time I was able to create the fibula guide almost completely in Slicer. I only made some small modifications in Blender.
The problem was that the boolean didn't work perfectly and parts of the miter boxes intruded through the guide base.

I have set the "Bigger miter box distance to fibula" to 0.
This is to avoid too large gaps between the miter boxes and surgical guide base.

The other option would be to make the surgical guide base thicker, but that would just be a waste of material.

I also tried to create the mandibula cut guide in Slicer, but gave up on it when I tried to create a tube that joins the cutting guides.

Right mandibula surgical guide base.
Looking at this image I'm thinking that it would be nice if the scene was less crowded. Would it be possible to only show the manipulator arrows for the segment that is active. By default everything would be inactive and when one clicked on a fibula segment the manipulator would appear.

I don't know if it's because I don't know how to manipulate fiducials in slicers or it's a feature - I was not able to place the fiducials where I wanted them to be. My expectation was that the fiducials would move only in a plane that is parallel to the view. That's not the case. When I moved the fiducial up, down, left or right, it also moved closer or farther away in respect to my current view. I tried for 15 minutes and then made the tube in Blender.

The final result.

I printed this guide with the new resin printer we got. The saw blade thickness is about 0,2 mm. As the resin printer is much more accurate in replicating small details the 1 mm pocket for the saw blade proved to be too big. In the next case I used a 0,2 mm pocket. With the 0,2 mm clearance it gave a nice fit.

[mauigna06]

Hi Martin. Thank you for keep logging your use cases. They are really useful. Please continue doing it.

The problem was that the boolean didn't work perfectly and parts of the miter boxes intruded through the guide base.
I have set the "Bigger miter box distance to fibula" to 0.
This is to avoid too large gaps between the miter boxes and surgical guide base.

This is not a problem of the boolean operations. This happens because "Bigger miter box distance to fibula" is zero. It gets the boxes nearer to the bone but not far enough to not touch some of the fibula. The intrusion problem would be avoided with an iterative positioning algorithm to position the boxes just a tenth of a millimeter away from the fibula or using a greater "Bigger miter box distance to fibula" (e.g. 3mm). Another option to correct this would be to allow a boolean difference with the fibula as last boolean operation when creating the fibula surgical guide. Do you have any comments on this @lassoan?
Neither of these two options would solve the large gaps between the miter boxes and surgical guide base. This can be solved using 3D sphere paint effect (with hollow effect extrusion of the fibula as editable area) to fill the gaps in the segment editor of the surgical guide base segment.

Would it be possible to only show the manipulator arrows for the segment that is active. By default everything would be inactive and when one clicked on a fibula segment the manipulator would appear.

@lassoan do you have any idea regarding this?

---

I don't know if it's because I don't know how to manipulate fiducials in slicers or it's a feature - I was not able to place the fiducials where I wanted them to be. My expectation was that the fiducials would move only in a plane that is parallel to the view. That's not the case. When I moved the fiducial up, down, left or right, it also moved closer or farther away in respect to my current view. I tried for 15 minutes and then made the tube in Blender.

Please try the suggestion @lassoan gave you:

You can choose between moving the points in the view plane or snap to visible surface in Markups module: Display / Advanced / 3D display -> unconstrained (moves in view plane) or snap to visible surface.

If you choose "snap to visible surface" and you create a model around the point and the model's Selectable property is True, then indeed, it becomes impossible to place the curve, because it will just run into the camera.

[lassoan]

Another option to correct this would be to allow a boolean difference with the fibula as last boolean operation when creating the fibula surgical guide. Do you have any comments on this @lassoan?

Sounds like a good idea.

Would it be possible to only show the manipulator arrows for the segment that is active. By default everything would be inactive and when one clicked on a fibula segment the manipulator would appear.
@lassoan do you have any idea regarding this?

I agree that the view is very busy. You could have a plane selector in the left panel and you would highlight that selected plane with a different color and you would only display arrows for that plane.

We have been discussing with @Sunderlandkyl about implementing a "selected" state for markups (for example when left-clicking on the markup) and displaying more controls (such as translate/rotate arrows) only for the selected markup, but this will not be available in the near future (this is not that easy because there are multiple views and you need to interact with multiple markups at the same time when you are in virtual reality).

Please try the suggestion @lassoan gave you: You can choose between moving the points in the view plane or snap to visible surface in Markups module: Display / Advanced / 3D display -> unconstrained (moves in view plane) or snap to visible surface.

Yes, this is exactly what you need to do. @mauigna06 I would recommend to add better support for this in the module and set the snap mode automatically. No need for using the "Markups to model" module, you can just ask the user to draw a curve and you can very easily generate a tube from it using vtkTubeFilter.

[mrtig]

Please try the suggestion @lassoan gave you: You can choose between moving the points in the view plane or snap to visible surface in Markups module: Display / Advanced / 3D display -> unconstrained (moves in view plane) or snap to visible surface.

Sorry for repeating this. I couldn't edit the original post anymore. So I copied case 5 and added the photos.

[mrtig]

Case 6

I'm not sure I have much to share about this case. Like the last case I was able to make the fibula guide almost completely in Slicer, but corrected the miterbox and fibula intersection in Blender. I also created the mandibula guide base surfaces in Slicer, but finished the guide in Blender.

I noticed that there is still some interaction between the first and last cut planes.

"Fix cut goes through mandible twice" didn't solve the issue. I later removed the piece in Blender.
Also, now I remembered that in this and in the previous case the "Create a 3D model of the reconstruction for 3D printing" button didn't work. It created a model in the data tree, but the model was empty.

So now I used a slot width of 0,2 mm and screw hole radius of 1,1. That gave a nice result. But unfortunately the problem with the resin I'm using is that it is quite brittle. The guide shattered when the oscillating blade hit the side of the miter box.

Also, in the previous case I used a very large diameter connecting tube between the mandibula surgical guide base surfaces. The surgeons had the idea that they could use it to fixate the right and left side of the mandibula together. They forgot that they would have to install a plate over the area that this guide covered. So it had to be removed anyway. Maybe it would be possible if I left a slot for the metal plate and moved the connecting tube towards the maxilla. I think this kind of a fixation would make their job easier and the result more accurate. Because it would be just the matter of putting the guide on the mandibula, making the cuts, inserting the fibula segments, putting a plate over everything and finally removing the guide..

[mauigna06]

I noticed that there is still some interaction between the first and last cut planes.
"Fix cut goes through mandible twice" didn't solve the issue. I later removed the piece in Blender.

Yes. "Fix cut goes through mandible twice" has a very simple implementation and only works if the bone segments are on only one side of the sagittal half of the skull scalar volume.

Also, now I remembered that in this and in the previous case the "Create a 3D model of the reconstruction for 3D printing" button didn't work. It created a model in the data tree, but the model was empty.

This happens when boolean operations fail. I would need to have a scene where this happens to make some tests and try to solve the problem. At least with the test data we have this feature works well.

So now I used a slot width of 0,2 mm and screw hole radius of 1,1. That gave a nice result. But unfortunately the problem with the resin I'm using is that it is quite brittle. The guide shattered when the oscillating blade hit the side of the miter box.

Maybe filling the gap with paint effect like I mentioned on the previous comment would help with this. Also, making the slotWall bigger may also improve the situation.

Maybe it would be possible if I left a slot for the metal plate and moved the connecting tube towards the maxilla. I think this kind of a fixation would make their job easier and the result more accurate. Because it would be just the matter of putting the guide on the mandibula, making the cuts, inserting the fibula segments, putting a plate over everything and finally removing the guide..

This comments are valuable. Please tell us if you could do it next time.

[mrtig]

I could share the scene, but I would need to anonymize it first. Is there a simple way for removing all the patient information?

[mauigna06]

As far as I know only the patient and the study folders save DICOM tags (like patientName, birthDate, etc) when a series is loaded from the DICOM database. ScalarVolumes do not contain patient information, only images. So it should be safe to anonimaze the data doing this:

Please upload a full scene of the virtual surgical planning completed so I can test why this feature returns an empty model.

[mrtig]

@mauigna06 Could you give me your email? I would like to send you the link privately.

[mauigna06]

Also, now I remembered that in this and in the previous case the "Create a 3D model of the reconstruction for 3D printing" button didn't work. It created a model in the data tree, but the model was empty.

@mrtig The problem is that decimated models contain non-manifold edges that brokens the boolean operations of Union between reconstructed fibula pieces and the resected mandible. To solve it I autocheck the use of non-decimated bone models and recalculate the resulting reconstruction ("Update fibula planes..." button) before making the reconstruction model. The changes are available on DentalImplantsBranch. Now clicking on "Create a 3D model of the reconstruction for 3D printing" will give a valid model.

[mauigna06]

@mrtig Can you add some quotes of the surgeons about what they said about the surgery final result (e.g. about accuracy)?

[mrtig]

@mauigna06 They usually give me feedback if something is not good. I think the accuracy has been good. With a previous guide there was inaccuracy because I made the miter box too wide. Most of the feedback has been about the design of the guides. The surgery time is decreased, but I don't have any figures to offer.

[mrtig]

Hi,

There was another surgery yesterday. This time a quite simple one - only one segment of the fibula was used in the reconstruction.
I made the fibula guide entirely in Slicer, but used Blender to make the mandibula guide.


I think I have got quite good settings for miter boxes The blade fits snuggly into the box and the box isn't too high.

This time I also attended the surgery and took some photos.


The fit was good. There was a maybe 0,5 mm gap between the fibula piece and mandibula. I'm thinking this might because of the clearance setting. It's set to 0,2 mm. @mauigna06 Could you clarify how this setting works?

[mauigna06]

The fit was good. There was a maybe 0,5 mm cap between the fibula piece and mandibula. I'm thinking this might because of the clearance setting. It's set to 0,2 mm.

Do you mean cap or gap? The fibula piece was a little longer or a little shorter than needed?

[mrtig]

Yes, sorry, I meant gap. The fibula piece was a little bit shorter. The length was good, but not perfect.

[mauigna06]

Supposing you did all the surgicalGuides on BoneReconstructionPlanner (guides with slots for the saw to pass through).

Different scenerarios can happen depending where the saw is positioned according to the guideSlot (considering that the surgicalGuide was printed with an ideal 3d-printer so no resolution related errors). The main information to analyse this is to see the code of the box creation to know the box's size and the translation that is applied to it (only the zAxis is important):

SlicerBoneReconstructionPlanner/BoneReconstructionPlanner/BoneReconstructionPlanner.py

Lines 3065 to 3068 in c70a742

	sawBoxWidth = sawBoxSlotWidth+2*clearanceFitPrintingTolerance
	sawBoxLength = sawBoxSlotLength
	sawBoxHeight = 70
	sawBoxModel = createBox(sawBoxLength,sawBoxHeight,sawBoxWidth,sawBoxName)

SlicerBoneReconstructionPlanner/BoneReconstructionPlanner/BoneReconstructionPlanner.py

Lines 3149 to 3157 in c70a742

	if i == 0:
	sawBoxAxisXTranslation = 0
	sawBoxAxisYTranslation = biggerSawBoxHeight/2+biggerSawBoxDistanceToMandible
	sawBoxAxisZTranslation = sawBoxSlotWidth/2
	else:
	sawBoxAxisXTranslation = 0
	sawBoxAxisYTranslation = biggerSawBoxHeight/2+biggerSawBoxDistanceToMandible
	sawBoxAxisZTranslation = -sawBoxSlotWidth/2
	sawBoxAxisOrigin = pointOfIntersection + sawBoxAxisX*sawBoxAxisXTranslation + sawBoxAxisY*sawBoxAxisYTranslation + sawBoxAxisZ*sawBoxAxisZTranslation

SlicerBoneReconstructionPlanner/BoneReconstructionPlanner/BoneReconstructionPlanner.py

Lines 2463 to 2466 in c70a742

	miterBoxWidth = miterBoxSlotWidth+2*clearanceFitPrintingTolerance
	miterBoxLength = miterBoxSlotLength
	miterBoxHeight = 70
	miterBoxModel = createBox(miterBoxLength,miterBoxHeight,miterBoxWidth,miterBoxName)

SlicerBoneReconstructionPlanner/BoneReconstructionPlanner/BoneReconstructionPlanner.py

Lines 2547 to 2557 in c70a742

	if i%2 == 0:
	miterBoxAxisXTranslation = 0
	miterBoxAxisYTranslation = biggerMiterBoxHeight/2+deltaMiterBoxAxisY+biggerMiterBoxDistanceToFibula/cosOfRotatedMiterBoxAxisYAndMiterBoxAxisY
	miterBoxAxisZTranslation = -miterBoxSlotWidth/2
	else:
	miterBoxAxisXTranslation = 0
	miterBoxAxisYTranslation = biggerMiterBoxHeight/2+deltaMiterBoxAxisY+biggerMiterBoxDistanceToFibula/cosOfRotatedMiterBoxAxisYAndMiterBoxAxisY
	miterBoxAxisZTranslation = miterBoxSlotWidth/2
	miterBoxOrigin = pointOfIntersection + miterBoxAxisX*miterBoxAxisXTranslation + miterBoxAxisY*miterBoxAxisYTranslation + miterBoxAxisZ*miterBoxAxisZTranslation
	miterBoxToWorldChangeOfFrameMatrix = self.getAxes1ToWorldChangeOfFrameMatrix(miterBoxAxisX, miterBoxAxisY, miterBoxAxisZ, miterBoxOrigin)

With the current code (and a sawBladeThickness exactly equal to the boxSlotWidth) these scenarios can happen:

• Benign case (similar to yours but with less length change):
  mandibleResectionLength = distanceBetweenFirstAndLastMandiblePlanes +2clearenceFitPrintingTolerance
  fibulaSegmentLength = distanceBetweenAdjacentFibulaPlanes -2clearanceFitPrintingTolerance
  (Because there is only one segment distanceBetweenFirstAndLastMandiblePlanes equals distanceBetweenAdjacentFibulaPlanes)
  The mandibleResectionLength is a little longer and the fibulaSegmentLength is a little shorter but the reconstruction can be done anyway

• Worst case:
  mandibleResectionLength = distanceBetweenFirstAndLastMandiblePlanes -2clearenceFitPrintingTolerance
  fibulaSegmentLength = distanceBetweenAdjacentFibulaPlanes +2clearenceFitPrintingTolerance
  (Because there is only one segment distanceBetweenFirstAndLastMandiblePlanes equals distanceBetweenAdjacentFibulaPlanes)
  The mandibleResectionLength is a little shorter and the fibulaSegmentLength is a little longer, probably the surgeon will need to abrade the fibulaPiece a little bit to make it fit.

The changes in length are very small if the clearenceFitPrintingTolerance is small. So the benign case and the worst case (and the cases in between them) shouldn't be considered negative.

If the math to calculate mandibleResectionLength and fibulaSegmentLength would be more rigorous the error between them would be even a little smaller.

@mrtig I hope this gives some clarity

[lassoan]

Thanks a lot for sharing these results. Probably the best would be to optimize the parameters by cutting up a couple of 3D-printed (or sawbones) fibula phantoms with the saw that is used in surgeries, with cutting guides created with different clearance/tolerance parameters.

[mauigna06]

@mrtig Thank you for letting us know about these surgeries. Please keep doing this kind of documentation. It is great

[mrtig]

Thanks a lot for sharing these results. Probably the best would be to optimize the parameters by cutting up a couple of 3D-printed (or sawbones) fibula phantoms with the saw that is used in surgeries, with cutting guides created with different clearance/tolerance parameters.

The small gap could have been caused by many factors. The cuts are made by a human. The mandibula guide holding the two pieces together could have stretched a bit. I will try to investigate with the 3D printed mandibles and fibulas.

The next thing I want to try is if it’s possible to install the metal plate without removing the mandibula guide. This should further increase the accuracy and simplify the workflow.

[mrtig]

There was another surgery a few days ago. I'm not sure I have much to add in terms of using the module. I'll post some photos and screenshots anyway.

Almost the entire left side was removed.

Used the same settings as with the previous fibula guide. I think the slots could be 0.1 mm wider. The fit of the saw blade was quite tight.


Designed the mandibula guides in Blender like previously. The one guide surface seems to be good in case of the mandibula. Space is often tight.

I used Formlabs Biomed Clear resin for printing the guides. Previously I used Dental SG resin and it was prone to shattering on saw blade impact. The fibula guide didn't shatter this time, but a chunk came off the mandibula guide. Anyway, not a big problem. need to make mandibula guide a bit thicker.

And a few photos of the surgery:


Unfortunately I didn't get any good photos of the mandibula guide.

[mauigna06]

Hi @mrtig. Thanks for keep logging these use cases. Please keep doing it. Good photos BTW.

need to make mandibula guide a bit thicker.

What was the thickness you used this time for the mandibleGuide?

Overall, by your use cases we can conclude that virtualSurgicalPlanning is really accurate and that surgicalGuides digital designs (miterBoxes and sawBoxes) are very good

[mrtig]

The thickness was 2mm, but I think the issue is a bit more complex than just thickness.

The accuracy is good.

[mrtig]

Next case
I discovered a bug. I have seen it before, but I think I have forgotten about it.
Sometimes there's a gap between the segments, it's about 0.3 mm.

Moving the cutting planes a bit will usually make the gap dissapear.

@mauigna06, Could you explain what is the purpose of the mandibular curve?


And the mandibula guide created in Blender like in previous cases.

[cmfsx]

Hi @mrtig when you have some time can you please share the video of the modifications you do with the Blender ?

Thanks

[mrtig]

Hi @cmfsx , currently I didn't really do any modifications as such. I created the mandibula guide using the resected mandibula model.
I can make a video of the process when I have some time. I'm a novice when it comes to using Blender so the methods I use are probably far from optimal.

[mauigna06]

Hi @mrtig. Thank you for logging these cases. Please keep doing it (add photos if it is possible)

I discovered a bug. I have seen it before, but I think I have forgotten about it.
Sometimes there's a gap between the segments, it's about 0.3 mm.

It would be good if you upload a scene where that problem appears (if you don't want to share the link here, send it to me via mail)

Moving the cutting planes a bit will usually make the gap dissapear.

That's a clue. I will be able to research this further with a reproducible example.

@mauigna06, Could you explain what is the purpose of the mandibular curve?

The mandibular curve is used to add mandible planes easily. It is also used to know if the mandible planes where created in curve-crescent or curve-decrescent direction (needed to create the fibula pieces models and resected mandible model). And it is used to create the mandible sawBoxes.

And the mandibula guide created in Blender like in previous cases.

@cmfsx, @mrtig would you like to create mandible resection guides like this (without slits, just use the wall of the resectionGuide to guide the saw)? I could add another option to create the sawBoxes like this.

[mauigna06]

Hi @mrtig.

I think the problem is, that in your BoneReconstructionPlanner version, Fibula Planes origins are being saved as interger arrays and that causes a submillimeter error. This bug have been fixed some months ago.
You can check by yourself if the origin position of the fibulaPlanes is right.
For example, in your scene file you I execute:

origin = [0,0,0]
getNode('FibulaPlane1_B').GetOrigin(origin)
origin

And this is printed:
[128.0, 39.00000000000001, 33.99999999999999]
As you can see these are all approximatelly integer values. That's wrong.

I just click 'Update fibula planes...' with the last code of the main branch and the gap visibly disappears.
Then I check the new origin of the fibulaPlane:

origin = [0,0,0]
getNode('FibulaPlane1_B').GetOrigin(origin)
origin

And this is printed:
[128.79677617998559, 39.83372026712204, 34.48418583825979]
All the values are real so that's correct.

To get that bug fixed you can do these things:

• download the stable slicer release and download BoneReconstructionPlanner from the Extension Manager (this will download the latest code).
• download the latest preview release and download BoneReconstructionPlanner from the Extension Manager (this will download the latest code).
• option to keep using your current slicer version: download the last code from the main branch and add it to the module paths of Slicer.

Let me know if with the update you find the problem solved.

[mrtig]

Hi,

It seems updating the software resolved the issue. I should have tried it beforehand.
Anyway, today was another surgery. I attended it briefly, but didn't take any photos.

The patient had a mandible fracture and the ends of the bone were on top of each other.

It took me some time to figure out how to best restore the shape of the mandible as it was before the fracture.

I exported the mandible to Blender and then cut it into two and rotated the pieces until it looked good.
Then I imported the model back to Slicer and converted it to a segmentation node.
Now I could use the model in Slicer to do the planning.

Later I used an ICP registration module in Blender to get the resected mandible pieces into their original positions and constructed the mandible guide in Blender as usual.

I also had to do the boolean operations for the fibula guide in Blender.

And for some reason creating the screw holes also didn't work in Slicer.

Traceback (most recent call last):
  File "C:/Users/user/AppData/Local/NA-MIC/Slicer 4.13.0-2021-10-21/NA-MIC/Extensions-30326/BoneReconstructionPlanner/lib/Slicer-4.13/qt-scripted-modules/BoneReconstructionPlanner.py", line 614, in onCreateCylindersFromFiducialListAndSurgicalGuideBaseButton
    self.logic.createCylindersFromFiducialListAndFibulaSurgicalGuideBase()
  File "C:/Users/user/AppData/Local/NA-MIC/Slicer 4.13.0-2021-10-21/NA-MIC/Extensions-30326/BoneReconstructionPlanner/lib/Slicer-4.13/qt-scripted-modules/BoneReconstructionPlanner.py", line 2613, in createCylindersFromFiducialListAndFibulaSurgicalGuideBase
    cylinderModel = createCylinder(fibulaScrewHoleCylinderRadius, "cylinder%d" % i)
  File "C:\Users\user\AppData\Local\NA-MIC\Slicer 4.13.0-2021-10-21\NA-MIC\Extensions-30326\BoneReconstructionPlanner\lib\Slicer-4.13\qt-scripted-modules\BRPLib\helperFunctions.py", line 194, in createCylinder
    cylinder.SetAttribute('height',str(H))
NameError: name 'H' is not defined

[mauigna06]

Hi Martin. Thank you for keep logging the cases. Please continue doing it.

I'll try to improve BRP so you can do all the workflow inside it. Maybe with a simplified bridge creation workflow it would be better. And I think that is the biggest inconvinience for you. What do you think?

Great idea to cut the mandible and reposition the pieces because the mandible was not symmetric before the VSP.

ICP registration is also possible on Slicer. And partial registration (like weighted-paint ICP in Blender) is possible in Slicer since two weeks ago thanks to the new SelectionByPoints tool @lassoan and me created for the DynamicModeler. And it is better than Blender because you have access to the transform nodes (and you could invert them for example).
@lassoan maybe this needs a video showing it exists. I'll do it and send you a link. Give me a few days.

I also had to do the boolean operations for the fibula guide in Blender.

You need greater than zero biggerMitterBoxDistanceToFibula to avoid this (this parameter should be big enough, e.g. 3mm, to avoid the mitterBoxes collisioning with the fibula). This will be solved with an iterative repositioning algorithm I will implement in the future (and this manual parameter will exist no more) as this can be the cause of a faulty guide. @cmfsx did you explain the importance of this parameter in the tutorial?

And for some reason creating the screw holes also didn't work in Slicer.

This bug was reported a week ago and it is solved in latest preview release of Slicer. Latest preview releases should be used because boolean operations are not possible in the stable release. And to update the code of the extension, you need to download and install the latest preview release of that day and install BRP. This is inconvenient for users. I may try to correct the bugs of the boolean operations module so the stable release can be used.

This bug shows the need of automatic tests for a whole workflow. @lassoan do you think it would be possible to implement a test of a full workflow in python (till both guides are created)?

[lassoan]

Yes, creating an automated test of the full workflow is competent doable. You can implement it at the logic level (call all the logic functions as they would be called during manual workflow) or at the GUI level (simulate GUI events) or a mix of both.

A video of the new surface paint feature would be awesome (accompanied by a Slicer forum and Twitter post). Without these users have no chance to discover these new features.

[mrtig]

You need greater than zero biggerMitterBoxDistanceToFibula to avoid this (this parameter should be big enough, e.g. 3mm, to avoid the mitterBoxes collisioning with the
fibula).

That would cause me to have a gap between the miterboxes and the guide base. The miter boxes and the base need to be joined firmly.

My suggestion would be make the miter boxes go through the fibula and just substract the fibula. Might not be so easy to make that work reliably.

I’m usually in a hurry so I use what I’m most comkfortable with. Rotating the mandibula pieces around the condyles is very easy in Blender - set rotation center to the condyle, rotate a bit around one axis and then the other until everything looks good. At the moment I don’t even have an idea on how to do it in Slicer.

[mauigna06]

That would cause me to have a gap between the miterboxes and the guide base. The miter boxes and the base need to be joined firmly.

This is because you use a narrow base, but I understand surgical guide resin is expensive.
The situation will improve when I have the mitterBoxesImproveContacts algorithm (based on collision detection with the fibula) working. But for such narrow guideBases, this may not be good enough.

The solution is to add another step to your surgicalGuideBases creation workflow. When you have the guideBase segment cut with scissors the way want, do something more:

• Copy the fibula segment to a new segment called 'hollowMask'
• Use hollow 25mm in 'hollowMask'.
• Select the guideBase segment, select paint tool, enable 3D sphere and edit in 3D view options and select in the 'Editable area' 'Inside hollowMask'.
• Paint over the guideBase where the gaps are (between the mitterBoxes and the guideBase). It will paint without disturbing the surface of the guideBase that touches the bone. I suggest a 12mm diameter for the paint brush

Try it and let me know if it solved your problem. You should be able to use now the parameter biggerMitterBoxDistanceToFibula greater than zero

[mrtig]

The resin is not so expensive, but I just can't justify to myself making the guide base 5+ mm thick.
I will try your workflow when I get some time.

[mauigna06]

In order to solve the problem of not-good-contact between biggerMiterBoxes and the fibulaGuideBase, I hand-crafted an adapted biggerMiterBox creation algorithm with lower faces that follow the fibulaLine like this:

Thus, these adapted boxes have better contact with the guideBase.
I yet have to add a iterativeContactOptimization algorithm for the adaptedBoxes so that the biggerMitterBoxDistanceToFibula parameter use is more accurate (i.e. the adapted boxes almost don't touch the fibula with biggerMitterBoxDistanceToFibula equal to 0mm).

@cmfsx, @mrtig would you be available to try this next week? If it works well with 2 or 3 plans of each of you I'll push it to the main branch. This should be easy for you because I think you can just load a previous planning scene and only click "Create miterBoxes" again and that's it.
The new branch will be available soon.

Should the adaptedBoxes be optional? Or just overwrite the default biggerMiterBoxes creation?

EDIT: here is the new branch, https://github.com/SlicerIGT/SlicerBoneReconstructionPlanner/tree/adaptedBiggerMiterBoxes

[mrtig]

That looks great. Sorry for not replying sooner. I will try to take a look at it when I have more time.

[mrtig]

I tried the the branch. I set the slot height to 7 mm. For some reason the slot height varies quite a lot.

[mrtig]

Another case. The surgery was already two weeks ago, but I just didn't have time to document it.

This time I tried painting the fibula guide base so it would meet with the saw boxes. It worked well.
I also tried making the mandibula guides in Slicer.
The first issue for me was that the saw box near the ramus should be longer. I guess a short one would also work because the upper part of the cut doesn't have to be exact. Anyway, I decided to make the box near the ramus longer. But it turned out during the surgery that I had made it too long so the surgeons used the old style guide I had also prepared.


Some photos form the surgery

[mauigna06]

Hi @mrtig. Thank you for sharing these cases, please keep doing it.

[mrtig]

Next case. I didn't attend the surgery so no photos.

First I tried without the "Use more exact version of the positioning algorithm" and the corner didn't align very well. Then I checked it and repositioned the segments. The result is on the image above.
I even think this option could be removed or at least the more exact algorithm should be the default. I didn't really notice it being any slower.

I also tried to use the BiggerMiterBoxes branch. For me the problem was that the boxes were not at the same height. So I reverted back to the main branch.

[mauigna06]

Hi @mrtig. Sorry for not answering earlier, I have been very busy.

Thanks a lot for keep logging these cases. It's really a great history. Please don't stop doing this.

The 'more exact planning' takes a little bit more time depending on the number of segments (more segments take more time) than normal planning. The difference of most exact planning with normal planning is that more exact planning creates a new fibulaFrame (with it's own centered fibulaLine that goes along the corresponding segment only) so this accounts for the deviations that the real anatomical axis of the fibula has. The result of this is that sometimes you can have better reconstruction results (because the created transforms are more accurate), but it not guaranteed. That's why it's not a default option.

For adaptedMiterBoxes, how would you like to measure it's height (for example, one way could be the normal distance to the guideBase) so that it's equal on all the miterBoxes? If you can show a picture of what you would like I can try programming it.

[mrtig]

For the adaptedMiterBoxes the behaviour should be the same as in the main branch, but it should just be possible to place them on the surface of the bone. I don't know how you have solved this problem algorithmically.
My idea would be that each miterbox would be moved along the cut plane until it become into contact with the bone and that would be the 0 for that particular miterbox.

[mrtig]

2 new cases this time. I didn't have time last week to write about the previous case.
First case
I had some trouble with the more exact version of the positioning algorithm. It produced better results, but adjustment was difficult because the segments rotated unexpectedly.

3 segments. One of the surgeons told me afterwards that it would've been better to make the reconstruction with 2 segments. Too much hassle with the small segments.
Also, the segments were a bit shorter than planned. Discussing with one of the surgeons he thought it might have been because the contact between the guide and the fibula was less than ideal.

Second case

This time I decided to create a 0.5 mm offset for the guide base. I'm not sure sure if that will change anything. I think this needs some trial and error to figure out how much soft tissue is left on the fibula.
As I had already finished the guide model I did the offset in Blender. I just extruded the fibula by 0.5 mm and did used the boolean difference modifier.
It would have been possible to also do it in Slicer, but I would've had to increase the volume resolution.
The surgery was supposed to be this week, but was postponed.
I have also been thinking about how it would be possible to attach the plate without removing the mandibula cut guide. It might be possible if the lower part of the mandibula guide has a gap into which the plate would fit.

[lassoan]

Some clearance always have to be added, not just because of the soft tissue residuals but also because of 3D printing inaccuracy. There is a nice and short related article about fits of 3D-printed parts: https://3dchimera.com/blogs/connecting-the-dots/3d-printing-tolerances-fits

[mrtig]

I have used a 0.5 mm offset on the latest guide base surfaces. It has worked quite well I think.
I actually just recently prepared another case which I didn't have time to document. I'm on vacation now.

[mrtig]

Some clearance always have to be added, not just because of the soft tissue residuals but also because of 3D printing inaccuracy. There is a nice and short related article about fits of 3D-printed parts: https://3dchimera.com/blogs/connecting-the-dots/3d-printing-tolerances-fits

This mainly applies to the tolerances of 3D printed parts that need to fit together.
It's easy to find out what good fit for the saw blade is by printing a few pockets with different widths. It depends on the printing technology. The saw blade used here is 0.2mm. If I print using the SLA printer I make the pockets 0.3 mm. With the FDM printer the width needs to be 0.6 mm.

[mauigna06]

@mrtig I think I may need to explain this in the written tutorial. But it's not clear to me if you take into account the tolerance setting of BRP that changes the size of the slots slightly (we talked about this previously on this thread). Could you clarify what clearanceFitPrintingTolerance you use for each case? Or what is the combined strategy of selecting the slot size and the clearanceFitPrintingTolerance (if you have ever changed it from default)?

[mrtig]

@mauigna06 I have not taken the tolerance setting into account and I have not changed the default value. So the actual pockets are pocket width+tolerance.

[mrtig]

Hi,

It's been a while since my last post. The end of last year was quite busy for me. I think we have had 5-6 cases since my last post.
I think my methods have become more efficient. I'm just going to go over my workflow.

Segment the mandibula and fibula (including blood vessels). I use the blood vessels as a reference.

Cut the mandible into segments

Used 4 segments for a better match with the original shape of the mandible.

Made the fibula guide base using the hollow function in segment editor to get a 3mm shell

Adjusted the between space to 3 mm because otherwise there would be very thin walls at the intersection of the mitre boxes.

Then went to models and enabled visibility in the red view for the biggerMiterBox objects and filled the gaps between the miter boxes and surgical guide base

I had to select "Red" view for every model individually. Is there an option for doing it all at once?

That's pretty much my BRP workflow. I still find it easier to create the mandibula cut guide in Blender. In Blender I also extruded the mandibula and fibula models by 0.5 mm and then substracted them form the final guides to get a small offset for any soft tissue left on the bones.

This time the surgeons bent the plate according to the patients mandible.

1. After the surgeons had prepared the mandible they attached the plate to the mandible with a few screws and drilled all the other screw holes into the mandible.
2. Then they attached the mandibula guide and made the two cuts.
3. Then attached the bent plate to the resected mandible.
4. Finally put the fibula pieces in between the resected mandible and secured them to the plate with screws.

The end result was quite good. There were some differences when compared to the planned reconstruction, but overall it looked good and the pieces fit together well.

On a previous case we also tried the following:

1. Bent the plate according to the planned model
2. Marked the screw hole locations on the model
3. Printed the mandibula guide without screw holes
4. Put the guide on the mandible model and drilled the holes in the locations marked on the mandible model (the guide material is transparent)

I was hoping this kind of workflow would increase accuracy of the surgery and remove a step from the surgery workflow. But actually the effect was the opposite - the surgeons had trouble getting a good fit between the plate and the resected mandible. And the fibula pieces didn't fit so well either.
I guess transferring the hole locations form the model to the bone this way has too many inaccuracies that add up.

[mauigna06]

Hi,

It's been a while since my last post. The end of last year was quite busy for me. I think we have had 5-6 cases since my last post.
I think my methods have become more efficient.

Hi dear @mrtig, glad to hear from you again!

it's very nice that you have using BRP in your work

I'm just going to go over my workflow.
Segment the mandibula and fibula (including blood vessels). I use the blood vessels as a reference.

You can check how useful is the new feature of Inverse Mandible Reconstruction to check correspondance of neck-leg vessels positions. Could you please give feedback about it?

Adjusted the between space to 3 mm because otherwise there would be very thin walls at the intersection of the mitre boxes.

Could you please explain the problem further with a picture and create an issue for it?

Then went to models and enabled visibility in the red view for the biggerMiterBox objects and filled the gaps between the miter boxes and surgical guide base
image
I had to select "Red" view for every model individually. Is there an option for doing it all at once?

I think latest release already does this could you please check?

In Blender I also extruded the mandibula and fibula models by 0.5 mm and then substracted them form the final guides to get a small offset for any soft tissue left on the bones.

Could you please create an issue for this?

This time the surgeons bent the plate according to the patients mandible.
After the surgeons had prepared the mandible they attached the plate to the mandible with a few screws and drilled all the other screw holes into the mandible.
Then they attached the mandibula guide and made the two cuts.
Then attached the bent plate to the resected mandible.
Finally put the fibula pieces in between the resected mandible and secured them to the plate with screws.
The end result was quite good. There were some differences when compared to the planned reconstruction, but overall it looked good and the pieces fit together well.
On a previous case we also tried the following:
Bent the plate according to the planned model
Marked the screw hole locations on the model
Printed the mandibula guide without screw holes
Put the guide on the mandible model and drilled the holes in the locations marked on the mandible model (the guide material is transparent)
I was hoping this kind of workflow would increase accuracy of the surgery and remove a step from the surgery workflow. But actually the effect was the opposite - the surgeons had trouble getting a good fit between the plate and the resected mandible. And the fibula pieces didn't fit so well either.
I guess transferring the hole locations form the model to the bone this way has too many inaccuracies that add up.

I'm not sure I understood correctly. Could you please create an issue with a more visual description? Then I could give ideas or code something to ease the workflow and increase accuracy

Thank you

[mrtig]

You can check how useful is the new feature of Inverse Mandible Reconstruction to check correspondance of neck-leg vessels positions. Could you please give feedback about it?

I'll take a look when I get the next case. I don't update the software very often because I have to contact IT every time.

Could you please explain the problem further with a picture and create an issue for it?

I'm not really sure it's worth a new issue.

Number 1 is a feature that is connected to the guide only from the back side, number 2 is entirely cut from the guide.

Could you please create an issue for this?
This is just my preference to do it in Blender. The only way I know how to do it in Slicer is using the Crop module and scaling the slice spacing and then I could create an offset that is small enough. It would be nice if BRP had the option of adding an offset to the models. I'll make the issue.

I'm not sure I understood correctly. Could you please create an issue with a more visual description? Then I could give ideas or code something to ease the workflow and increase accuracy

I don't think this is related to the accuracy of BRP.
It's related to the accuracy of transferring the hole location between the mandible model and patient's mandible.
The plate was bent on the reconstructed mandible model and the hole locations were marked with a sharpie.

Then the mandibula guide was placed on the model and holes were drilled in the guide in the marked places.

The idea was to use the same holes for attaching the guide and the plate. Ideally this would help to maintain the position of the resected mandibula pieces in relation to each other.
But this method of transferring the hole locations is not accurate enough. The surgeons had a bit of trouble getting the plate to fit correctly on the patient.
Also, because all of the cuts are made by humans there will be some deviations from the guides. CNC accuracy is not possible.

[mrtig]

Another case. This time it was a case of radionecrosis. The patient had already had reconstructive surgery some years ago, but now had to have a new surgery. The jaw had become much smaller so I made a reconstruction that was something between his pre-surgery mandibula and the current state.

Green is the pre-surgery mandibula. Blue is the current state.

Red is something in-between.

I did the segmentation in Slicer, but did the necessary manipulations of the model in Blender and then imported the model as a segmentation back to Slicer.

The resected mandible

Fibula guide

I upgraded to the latest version of Slicer, but still had to enable the Red view manually for every bigger miterbox model to see them on the red slice.

[mauigna06]

Hi @mrtig

Thank you for your feedback.

upgraded to the latest version of Slicer, but still had to enable the Red view manually for every bigger miterbox model to see them on the red slice.

You can execute self-tests as shown on the video here and check that biggerMiterBoxes are indeed visible for the Red view.

I guess you just need to toggle the currentScalarVolume selector to update the display properties of the boxes. Please let me know if that works for you

[mrtig]

@mauigna06 after upating Slicer I could see the miter boxes in the red view. Thanks!

[mrtig]

Another case. I think this has been the most extensive so far. A large part of the mandible was removed.

The reconstruction:

Mandibula guide:

Fibula guide:

Surgery result:

The result was very good. The fibula pieces and the plate fit together very well. I think this was one of the fastest reconstructions I have seen - I mean the part where they fit the fibula pieces, plate and the resected mandible together. Usually there is some fiddling around to get everything to fit together, but this time it was very quick.

[mauigna06]

Hi @mrtig
Thanks a lot for the update, it looks very nice

[mrtig]

@mauigna06
Thank you for all your work and effort!

[mrtig]

There have been a few cases since my last post.

I tried something new this time. I made a guide for aligning the fibula segments.

The lower portion of the pieces sit in the guide and it has screw holes in the bottom for fixation.

The surgeons called it the "trough" and they really liked it.

[mrtig]

The trough holds the fibula pieces together so the reconstructed part can be inserted as one piece. It makes work a lot easier because before there always been a lot of fiddling to place all the segments and the plate in the correct positions.

I created the trough in Blender. So maybe it's not so useful to post an issue about it here.

[qferreira-1]

I have read research papers published by the Mayo Clinic utilizing something similar to what they called trays. I have provided this for one case which required 3 osteotomies and created it quickly in Meshmixer with a simple box primitive and imported the reconstructed mandible from Slicer in Meshmixer and made some simple boolean operations to create the tray. They used it similarly to @mrtig's surgery, by holding the fibula segments for plating or alignment as doing so during surgery can be tricky.

[mauigna06]

@mrtig, @qferreira-1

I could add a feature to allow easier joining of the fibula pieces, you would just need to draw a guidebase over the superior part of neomandible VSP and use the dentalImplants feature to create holes for kirschner wires.
Could you ask the doctors if they would like it? :)

[qferreira-1]

@mauigna06 : When we trialled this for a case, the physician said it was a unique idea for fibula placement; however, the practice changes from surgeon to surgeon. Our guy typically pre-bends the plate and we have an external fixation guide that holds the mandible in the occlusal position for the physician to install the plate. Once the plate is installed, they place the segments and fixate them. He did note that if we had a system for predictive screw holes, it could potentially become more useful.

[mauigna06]

Ok. Thanks for the feedback :)

[mrtig]

There have been a few surgeries since I last posted. The tray has been useful in some cases, but not all. In some cases there's been soft tissue on the lower side of the reconstruction. So it has not been possible to use the tray.
The tray was used quite successfully during the last surgery.

3D model

The tray is fixed to the underside of the reconstruction

The main advantage of using the tray is that it takes less time to get the all the segments into correct position and fix them using the metal plate.

A disadvantage is that it occludes the underside of the reconstruction. So if the fibula segments are not fully seated in the tray there might be some misalignment of the segments.

The workflow was following:

1. microsurgery of the blood vessels
2. attaching the resected mandible and the fibula segments to the tray
3. attaching the metal plate to the assembly that the tray is holding together

Also, here's the fibula guide:

I added the horseshoe shape reinforcements on the request of a surgeon, because the saw would often break the guide in the place where two pockets are very close to each other. I think this is only a problem with guides made of resin, because resin is much more brittle than thermoplastics.

[mauigna06]

Dear @mrtig

Thanks a lot for keeping us updated about your use of the software.
Your design changes ideas (like the 'tray' or the 'horseshoe') and the feedback are very useful

Please stay keeping us posted :)

[mrtig]

This time I helped to plan a maxilla reconstruction.

The process was the same as for mandibula, but I was only asked to create a model of the reconstruction and the fibula guide. The cut was made right between the front teeth so there was need for a maxilla cut guide.
Here's the reconstruction:

The reconstruction was used to bend the plate.

[mauigna06]

Hi @mrtig
Thanks for sharing :)