Digital Bite Accuracy: Hidden Scanner Errors and the AI Fixes

Dr. Melissa Seibert: Welcome to Dental Digest. This is a podcast with a mission of enabling you to stay on the cutting edge of evidence-based dentistry. I'm your host, Dr. Melissa Seibert. Today's episode is Part 1 of a two-part series with Dr. Mike Skramstad. Dr. Skramstad is one of those rare clinicians who lives at the intersection of clinical dentistry, digital workflows, and product development. He's a longtime CEREC and CAD/CAM educator, faculty at Spear Education, an alpha and beta tester for multiple major dental manufacturers, and someone who has spent decades stress-testing digital dentistry before most of us were even considering it.

In Part 1, we take a broad look at how digital dentistry actually functions in the real world — not marketing decks, but busy restorative practices. We talk about scanners, mills, CAD/CAM systems, and where the true limitations still exist, especially when it comes to occlusion, bite accuracy, and predictability. Dr. Skramstad shares why certain digital systems succeed where others quietly struggle and what dentists often misunderstand about accuracy and digital workflows. This episode is less about chasing the newest gadget and more about thinking critically about how digital tools integrate into clinical decision-making, team workflows, and patient care.

Dr. Seibert: With digital scanners, you can capture occlusion with remarkable repeatability, yet we still see discrepancies between mounted analog cases and digital bite registrations. What do you think most clinicians introduce error? Is it the capture, the alignment, or the interpretation?

Dr. Skramstad: Well, there's actually research on this. Research I presented at the Spear Summit. If you look at that, there's like eight to ten factors that go into the discrepancies on digital bites. It could be how many teeth you scan. It could be the scanner itself. It could be — are there edentulous spaces between the teeth? If you go from tooth to a big edentulous space to a tooth, that would affect it. It could affect tooth mobility. It depends on how hard the patient bites. Occlusion is probably the biggest weakness of intraoral scanners right now. It's hard to get predictability.

If you took six different scanners and you scan the exact same case, you're very likely going to get six different occlusions. They're all going to be pretty close. If you're doing a single crown, it's close enough. The research has shown that it's close enough and it's predictable and repeatable. But when you start getting into full-arch dentistry, when you start needing that accuracy of occlusion, that's when you start running into issues.

These are the reasons that things like BiteFinder — I don't know if you've heard of that — are coming to market. And there's another scanner that was just literally introduced yesterday from Shining 3D called, I believe, ELF. It's a brand-new scanner and they have AI within that scanner that's supposed to help re-articulate the bite using AI. So hopefully this is something AI is going to help us.

Dr. Seibert: Tell us a little bit more about BiteFinder. I remember you talking about it. I furiously took notes when you were talking about it.

Dr. Skramstad: BiteFinder is an AI company that was originally introduced to the labs, quite honestly, because the labs have it harder than we do. If you're in my office — where I do most of my dentistry in-house with an in-house lab — and my bite is off significantly, I'm probably going to be able to tell, because I've been in the mouth for an hour. The labs have a whole different problem: they're getting a case and they don't really have a way to tell, "Is this bite correct or not correct?" — which leads to tons of remakes.

BiteFinder is an AI program. You can use it on Google Chrome — it's web-based — or you can integrate it into ExoCAD. What it does is it looks at the teeth themselves: what do the teeth look like? What are the wear facets in the teeth? For instance, if this wear facet was here, how would that possibly get there? It runs it through their algorithm and tries to fit together or re-articulate the teeth using AI based on the way the teeth look themselves. It's surprisingly pretty accurate. There's research that's been done that shows it's more accurate. If you take the bites raw out of the scanner, then put all these different scanners into BiteFinder, it tends to bring them all to the same place.

Dr. Seibert: Walk us through practically — are you uploading the STL into BiteFinder and then sending it off to the lab? If you're doing the CEREC workflow, that's a closed-loop workflow, so you can't really use BiteFinder there.

Dr. Skramstad: There's two ways. If I'm sending something to the lab — let's say I'm scanning with the PrimeScan — I don't always do this, it depends on the dentistry and how precise it needs to be. For a bite splint, generally I feel pretty comfortable that even if it's off a little, I can adjust it. But if I'm doing something that needs a lot of precision, I'll scan with the PrimeScan, export the STLs to my desktop, then open Google Chrome and go to the BiteFinder site. BiteFinder sells bites in packages — say, 20 different bites for X dollars. You download or upload those scans into BiteFinder and run it through their algorithm. It honestly takes about 20 seconds. It will export the new re-articulated bite, and it also generates jaw motion data.

The same way most of these articulators — like in ExoCAD — have built-in articulators with average values. The problem with those articulators is the average values are only as good as your bite. If your bite is off, they're useless. BiteFinder gives you a new articulated bite and jaw motion data. I'd export that whole file to my laboratory to work with. You can also use PLY files if the lab prefers color.

Now, if I'm doing it myself in office, these bigger cases I'm not using CEREC for. I'm using ExoCAD. There's a little script you can download from BiteFinder that integrates directly into the dental database. Right when you open up your case, you can start BiteFinder and it'll re-articulate it directly within ExoCAD. You don't have to import and export things.

Dr. Seibert: Jaw motion tracking — yes or no? Is that something you have in your office?

Dr. Skramstad: I don't. We've talked a lot about this. The cream of the crop is ModJaw. It works — there's some pretty fascinating research, and we have it at Spear. But the problem in my mind with ModJaw is, as fantastic as it is, it's super expensive. Unless you're doing that type of dentistry predictably in your office, I don't see the value point. If you do, then 100%. If all I was doing was big prosthodontic cases, or if I was doing a lot of them, I kind of equate it to photogrammetry. If you're doing a lot of those cases, absolutely it makes sense. If not, then maybe not.

When it comes to ModJaw, it just depends on the type of dentistry you're doing. It is the — I'd say it's actually better than analog. Really accurate. My question is, are things like AI going to potentially do the same job over a period of time? That's when I talk to Dr. Darin Dichter at Spear — we're kind of the two digital guys that bounce things off each other. He's using ModJaw. I'm wondering if things like BiteFinder, as they evolve, maybe there'll come a point where you'll upload the CBCT into ModJaw and it'll all use AI to create a similar ecosystem. Time will tell.

Dr. Seibert: Speaking of scanners and using them for full-arch cases — would you scan a full-arch case or take an analog PVS impression?

Dr. Skramstad: I would always scan. I haven't taken a PVS impression in a long time. The only time I have PVS in my office is for relining dentures so I can scan them. Everything is pretty much scanned. I also have a laboratory I've been using for a long time — we have a very high level of trust. He knows what he needs, and I know what he needs. If something I send him doesn't work, he'll immediately get back to me. I always scan. I feel very comfortable with it.

Dr. Seibert: Spill the beans — what lab is it?

Dr. Skramstad: I have a couple. It depends on if you want removable. The lab I'm using for most of my veneers and big cases is called Westland Dental, up in the Minneapolis area. Shout out to Elliot — he's great.

Dr. Seibert: When it comes to full-arch cases, what do you do to improve your accuracy?

Dr. Skramstad: It depends on a lot. The most simple thing is to scan with articulating paper marks. Before you scan the upper jaw, have them go into articulating motion, or just in MIP, have them tap really hard. You'll see the blue marks in the teeth. Then scan your upper arch and see: do the proposed occlusal contacts in the software match the actual marks in the mouth? If they don't, the next step would be running it through BiteFinder to see if it does. Those are the two most common ways I would guarantee it.

It also depends on the scanner. I don't really play favorites because I have a lot of them and they're all great. This is just my personal observation, not official research — but I think Lumina probably does the best job of all scanners with occlusal accuracy. It's got weaknesses for sure, but if you want to scan a full-arch case and get the best occlusal accuracy, that's the best I've seen.

Dr. Seibert: This being the iTero scanner. Where's the PrimeScan?

Dr. Skramstad: It's good. The PrimeScan to me is best with teeth — it scans teeth the best. The Trios scans tissue the best. The scanner I choose depends on what I'm going to do. With the iTero Lumina — and this is just a theory — if you've seen it, it has a bigger tip, a big flat tip, and there are six cameras in there. My theory is the bigger the tip, the more data it captures at one time. It's a different scanning experience. It almost feels like you're scanning cloud pixel points. It was really hard for me to get used to because it doesn't pick up data the same way, but it picks up a lot at once.

I think where the inaccuracy comes from with scanners is when you have a smaller field of view, you have to scan a lot of different areas and stitch them together — that's what creates inaccuracy. It's not necessarily the buccal bite — it could be the way the model is constructed by scanning little pieces. The Lumina, since it's so big, that's probably the reason for higher accuracy. There's official research coming out focusing on scanning scan bodies. For full-arch fixed removable, the standard has been photogrammetry — but now you're starting to see intraoral fancy scan-ladder type things where you can scan with Shining 3D. I think True Abutment has one. You need certain scan bodies. iTero Lumina is claiming they can scan regular scan bodies with their scanner and get global accuracy matching photogrammetry. I've never done it, but that's what the paper says.

Dr. Seibert: Some things I do to improve accuracy: first, following manufacturer-specific scan guidelines, making sure it's reasonably dry, no excessive light shining, not constantly going over the same site. Would you say those rules still apply today?

Dr. Skramstad: I think so. They might be becoming less important than they used to be, but they're still important. The scanners have evolved so far. When we first started, this wasn't even a possibility. We started with quadrant scans and powdering teeth back in the old days. Now we're seeing scanners do full-arch scans in 15 seconds. It's crazy how far they've come.

I'm still a big believer in scan path. If you look closely, most manufacturers have a similar scan path. They're all fairly similar — there's a reason behind that. Capturing big chunks of data first, then filling in minute details later. When you get into scanning one tooth, then the next tooth, what you end up doing is scanning the same area over and over again. That inherently builds inaccuracy. The rule of thumb is: only scan as much as you need to get the job done.

Dr. Seibert: Owning a practice is still a business — we want to be efficient. Are you having your staff, especially with the CEREC workflow, do the scanning, designing, and restorations? I like a lot of precision; sometimes that means I like to control the process, but sometimes that's not profitable. Where do you stand?

Dr. Skramstad: It depends on the dentist in my office. Everybody does things differently. I do more of a hybrid workflow. If we're doing an implant guide — we take a CBCT, take a scan with the PrimeScan, merge it together, design and 3D print the guide — my assistants do everything. They take the CBCT, take the scan, design the surgical guide. I approve it and make any adjustments before they 3D print.

For bite splints, the assistants do all the scanning. For restorative work, it's hybrid. I'll anesthetize the patient. They usually scan the patient — upper or lower bite — while the patient's numbing up. Once they're done, I come in, I prep the tooth. Say it's a simple single crown — I prep, I take the final scan. I like to take the final scan and I like to do the design. I do all the designs and margination. Some dentists let their assistants do a lot of that and they approve it. I know dentists who don't even know how to run the software — their staff does everything. I'm way too picky for that. The final result is on me. Plus, honestly, I'm so much faster than they could be. From an efficiency standpoint, it takes me hardly any time to design a single crown. It makes no sense for them to do that — they can be doing something else.

Dr. Seibert: I'm very picky too, very particular. I'm trying to learn what's worth delegating and what's not.

Dr. Skramstad: When you're like that and you hire the right assistant — I've had the same assistant for 18 years. She's worse than I am. She's super picky. I'll be prepping and she'll say, "You need to smooth that angle. That's not smooth enough. Go over that margin." I've rubbed off on her. She has certain skills I don't have — when it comes to staining and glazing, she's far better than I could ever be. She has a talent for it. So I do delegate certain things.

Dr. Seibert: How did you find her?

Dr. Skramstad: Do you want to hear the story? When I was first starting an office, I was under the impression you had to hire people with experience. I have the complete opposite theory now — I want them right out of school, especially hygienists. When she applied, she was 19 or 20, just out of assisting school. I thought, "I'm never going to hire her. I don't need a young girl who doesn't know anything." But I looked at her resume, and at the bottom — one of the people who was a groomsman in my wedding was her reference. I called him. "Scott, what's going on with this? Who's this Jamie? Why didn't you tell me?" He said, "Oh yeah, I forgot to tell you about her. You should hire her. She's been running my cell phone stores since she was 15. She's amazing. She's the smartest person ever. In fact, I'm mad she went to assisting school and doesn't work for me anymore. Just hire her." So I interviewed her for 30 seconds, told her Scott said to hire her — "I guess you're hired." Complete luck on her end and mine.

Dr. Seibert: Do you feel like if they have the character and characteristics you're looking for, their experience level is almost negligible? You can also run into issues if they're the old crusty ones who say, "Such-and-such has been doing it this way forever."

Dr. Skramstad: Experience doesn't matter as much as how they fit into the cohesive environment of my office. I have 29 employees. There are two of us that are male — me and another younger dentist — and everybody else is female. My dynamic is: I can get along with anybody. But it's how do you get along with everybody else? You have to have a feel for that. When we have someone we think we'll hire, their second interview is to come have lunch with the whole staff. They decide. Most of the time it's: "Yeah, she's great." Once they said, "I don't think she's going to fit," and we didn't.

We've had more than one assistant come to us before they were even certified. They couldn't take x-rays. Just from watching the other assistants, being helpful, being part of the team, they learned — and now they're full great assistants. If you're patient and have the right staff that's patient, you can train people to be great. They don't have to come to you great. I almost prefer it. Sometimes when they come with a ton of experience, they're stuck in their ways and resistant to change. That's usually what creates issues. People open to changing and doing things a different way — because I do things quite a bit differently than some — are the people I tend to gravitate toward.