Realize Medical

XR Demystified for Surgeons

Where 3D changes the plan, and where it rarely does

Last reviewed October 2026

The short version: in the larger published studies, seeing a case in 3D changed the surgical plan in anywhere from about 3% to 75% of cases. The rate climbs with case complexity, from unusual congenital anatomy at the top to routine neurosurgical cases at the bottom, and it depends on what each study counted as a change. These studies count changed plans. They don’t show that the changed plans led to better outcomes.

3D model of a child's heart with complex congenital anatomy, seen from the front and left. The right ventricle is see-through blue with the red left ventricle behind it, and the orange aorta and violet pulmonary arteries both rise from the right side of the heart.
  • Right atrium
  • Right ventricle (see-through)
  • Pulmonary arteries
  • Left atrium
  • Left ventricle
  • Aorta and coronary arteries

Complex congenital heart (double-outlet right ventricle). Illustrative 3D model from a de-identified case, built in Elucis. Not from the cited studies.

Reading cross-sectional images is a hard-won skill, and for most cases the slices answer the question. 3D reconstruction isn’t new either; radiology has offered it for decades. The useful question is narrower: in which cases does seeing the anatomy in 3D change what the team plans to do?

The first guide in this series, When 3D helps, and when it doesn’t, covers what slices leave to the reader. This one looks at a single number: how often the plan changed.

What counts as a “plan change”

WHAT EACH STUDY COUNTED

Same question, five different yardsticks

LIPPERT 2024

Heart-team strategy, before and after a mixed-reality model

First meeting on standard imaging; second meeting with the model, a median of 19 days later. Moving from undecided to decided counted as a change.

Prospective · single centre · 50 children with complex congenital heart defects

PELIZZO 2025

Surgeon-reported change of approach after VR review

Scored retrospectively, after surgery, from the operating team’s feedback.

Retrospective · no control group · 50 VR models, children with malformations or tumours

BAKHUIS 2023

“Critical” changes to a segmentectomy plan

Three surgeons planned from CT, then reviewed the case in VR. Counted: a different segment, a smaller resection, or an extended one.

Prospective · 50 consecutive patients

SHIRK 2022

Surgeon-reported changes in the trial’s 3D-model arm

Mostly nerve-sparing and margin decisions. Patients in the other arm had no model, so had no plan change to count.

Randomized trial · 92 patients, 41 in the model arm

COLOMBO 2023

A change of surgical approach after a hologram

Viewed in theatre between anaesthetic induction and draping, when most of the plan is already set.

Prospective · 107 consecutive neurosurgical cases

Sources: 1, 2, 3, 4, 5

Each study compared a plan made from the usual images with the plan after the team reviewed a 3D model of the same patient. What counted as a change differs from study to study, and that matters as much as the percentage.

A change can be as large as a different operation, or as small as deciding between options the team had already discussed.

The range, from complex to routine

Five of the larger studies in the research library (40 or more patients each) show the spread. Smaller case series and surgeon surveys report higher rates, so a single headline number says little on its own.

At the top are children with complex congenital anatomy.21 In the middle, segmentectomy, where segment borders are hard to place on 2D CT,3 and robotic prostatectomy.4 At the bottom, routine neurosurgical cases.5

HOW OFTEN THE PLAN CHANGED

About 3% to 75%, depending on the cases

  1. Children with congenital malformations or tumours, Pelizzo 2025 · retrospective · 50 VR models: plan changed in 75%, count not given.
  2. Complex congenital heart defects (children), Lippert 2024 · prospective · 50 patients: plan changed in 68% (34 of 50).
  3. Lung segmentectomy, Bakhuis 2023 · prospective, consecutive · 50 patients: plan changed in 52% (26 of 50).
  4. Robotic prostatectomy, Shirk 2022 · randomized trial, model arm · 41 patients: plan changed in 32% (13 of 41).
  5. Routine neurosurgical cases (tumour, vascular, carotid), Colombo 2023 · prospective, consecutive · 107 cases: plan changed in 3% (3 of 107).
filled bar = count reported in the paperoutlined bar = percentage only, no count

Rates are not directly comparable: each study counted a different kind of change (see above). These studies didn’t use Elucis.

Sources: 1, 2, 3, 4, 5

Why complex cases change plans, and routine ones rarely do

EACH DOT IS ONE PATIENT

Two ends of the range, and the middle

COMPLEX CONGENITAL HEART · LIPPERT 2024

34 of 50

Strategy changed after the model

SEGMENTECTOMY · BAKHUIS 2023

26 of 50

Plan changed after VR review

  • Different segment 14%
  • More lung-sparing 10%
  • Extended resection 28%

ROUTINE NEUROSURGERY · COLOMBO 2023

3 of 107

Approach changed after the hologram

plan changed no change
Sources: 1, 3, 5

Where the plan was still open, there was more to change. Lippert and colleagues only enrolled children whose first heart-team meeting hadn’t settled on a plan, or had several on the table. After the model, the team’s strategy changed in 34 of 50.1

Where 2D makes the anatomy hard to place, plans shift. In Bakhuis and colleagues’ 50 consecutive segmentectomies, the plan changed in 52%: the tumour was placed in a different segment in 14%, a more lung-sparing resection was planned in 10%, and an extended resection in 28%.3

Where cases are routine and the plan is set, it rarely does. Colombo and colleagues collected every case their centre prepared with a hologram over a year: 107 consecutive tumour, vascular and carotid cases. The approach changed in 3.5

In between, the changes are often refinements. In Shirk and colleagues’ trial, surgeons changed the plan for 13 of the 41 patients in the model arm (32%), mostly decisions about nerve sparing and margin width rather than a different operation.4

3D model of a right lung seen from behind and to the side. The upper, middle and lower lobes are see-through in mint, lilac and sky blue; blue pulmonary arteries and red pulmonary veins branch down into the lower lobe, where a small amber tumour sits among the branches.
  • Upper lobe
  • Middle lobe
  • Lower lobe
  • Pulmonary arteries
  • Pulmonary veins
  • Tumour
  • Airway

Right lung with a tumour in the lower lobe. Illustrative 3D model from a de-identified case, built in Elucis. Not from the cited studies.

What these studies don’t show

  1. 1

    That the changed plans led to better outcomes

    The counts come from the people doing the planning. They show that the model changed a decision, not that the new decision was better. Only one of the five, Shirk and colleagues’ trial, was randomized and followed patients; its primary outcome, positive surgical margins, did not differ significantly between the groups.4

  2. 2

    The effect of the model on its own

    Four of the five had no comparison group for plan changes: the team planned first without the model, then with it. That design can’t separate the model from a second look. In Lippert and colleagues’ study the team’s plan changed between the two meetings in 68% of cases, while individual members changed their view after seeing the model in 39%.1

  3. 3

    How sure the top figure is

    Pelizzo and colleagues’ 75% comes from a retrospective review with no control group, scored after surgery, and the paper doesn’t give the count behind it.2

  4. 4

    Which display, or which product

    The studies used mixed-reality headsets, VR headsets and a smartphone 3D app. None compared displays, and none used Elucis.

Read the numbers as a guide to where 3D is most likely to change a plan, not as proof that it improves care. Every study, with its limits, is in the research library: /research/q/plan-change.

If you have a case like this

Unusual anatomy, a plan the team hasn’t settled, or a decision that hinges on how structures relate in 3D: these are the cases where the studies above saw the most plan changes. Slices stay the reference; a model is something to check against them.

Elucis is FDA-cleared (510(k) K220649)6 software for building 3D models from medical images and planning with them, used in conjunction with expert clinical judgement.

Have a case like this coming up? The Realize Medical team can walk through it with you in Elucis.

Walk through a caseSee all plan-change studies

Previous guide: When 3D helps, and when it doesn’t

References

  1. Lippert M, Dumont KA, Birkeland S, et al. Cardiac anatomic digital twins: findings from a single national centre. European heart journal. Digital health. 2024. PMID 39563912. https://doi.org/10.1093/ehjdh/ztae070
  2. Pelizzo G, Pierucci UM, Marinaro M, et al. Virtual Reality for Preoperative Planning and Education in Pediatric Surgery: Preliminary Results for the Treatment of Congenital Malformations and Tumors. World journal of surgery. 2025. PMID 40246557. https://doi.org/10.1002/wjs.12594
  3. Bakhuis W, Sadeghi AH, Moes I, et al. Essential Surgical Plan Modifications After Virtual Reality Planning in 50 Consecutive Segmentectomies. The Annals of thoracic surgery. 2023. PMID 36084694. https://doi.org/10.1016/j.athoracsur.2022.08.037
  4. Shirk JD, Reiter R, Wallen EM, et al. Effect of 3-Dimensional, Virtual Reality Models for Surgical Planning of Robotic Prostatectomy on Trifecta Outcomes: A Randomized Clinical Trial. The Journal of urology. 2022. PMID 35848770. https://doi.org/10.1097/JU.0000000000002719
  5. Colombo E, Regli L, Esposito G, et al. Mixed Reality for Cranial Neurosurgical Planning: A Single-Center Applicability Study With the First 107 Subsequent Holograms. Operative neurosurgery (Hagerstown, Md.). 2023. PMID 38156882. https://doi.org/10.1227/ons.0000000000001033
  6. US Food and Drug Administration. 510(k) K220649, Elucis (Realize Medical, Inc.). Decision 2023-01-17. https://www.accessdata.fda.gov/scripts/cdrh/cfdocs/cfpmn/pmn.cfm?ID=K220649

Percentages and counts are as reported in each paper; counts in brackets for Bakhuis are 14%, 10% and 28% of 50 patients.