A Digital Twin for Robotic Post Mortem Tissue Sampling using Virtual Reality

Alexander Schlaefer; Axel Heinemann; Benjamin Ondruschka; Kristina Allgoewer; Ludwig Bosse; Maximilian Neidhardt; Vidas Raudonis

REVIEW 3 major objections 5 minor 16 references

A Digital Twin for Robotic Post Mortem Tissue Sampling using Virtual Reality

T0 review · 3 major / 5 minor · reviewed 2026-08-05 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A virtual-reality digital twin lets pathologists plan and supervise robotic post mortem biopsies from a separate room; 132 insertions landed within 5.30 ± 3.25 mm off-axis.

desk verdict A solid integrated feasibility study: remote VR planning plus robotic post mortem biopsy works on cadavers, but the headline accuracy number is softer than it looks because excluded insertions and possible target displacement are both in play. read the letter →

arxiv 2509.02760 v2 pith:7TBP2QSS submitted 2025-09-02 cs.RO

Maximilian Neidhardt , Ludwig Bosse , Vidas Raudonis , Kristina Allgoewer , Axel Heinemann , Benjamin Ondruschka , Alexander Schlaefer This is my paper · ORCID

classification cs.RO

keywords virtualrealitydigitaltwinroboticneedleplacementpostmortembiopsyminimallyinvasivetissuesamplingCT-guidedinterventionteleoperatedrobothuman-robotinteraction

verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper reports a complete remote workflow for post mortem tissue sampling: a pathologist wearing a VR headset plans needle trajectories on CT-derived anatomy in a life-size digital twin of the intervention room, and a lightweight robot in the physically separate room executes the insertions. Across three cadavers of different body types, 132 needle insertions achieved a mean off-axis placement error of 5.30 ± 3.25 mm and a mean surface puncture error of 2.62 ± 1.76 mm, comparable to the authors' earlier robotic system. Histopathological examination confirmed the intended tissue in 65% of 46 extracted samples, with misses concentrated in small targets such as coronary arteries. Eleven forensic pathologists and ten non-physicians completed the planning task after a five-minute introduction; the two interaction modes that let the user grab and orient the needle in 3D were faster and rated more intuitive than a conventional triplanar screen-based mode. If these results hold, the system offers a practical, low-exposure alternative to manual ultrasound-guided PM biopsy, especially for large-scale systematic sampling.

What carries the argument

The load-bearing mechanism is the digital twin's single shared coordinate frame. A chain of rigid transformations—robot base, tracking camera, phantom on the cadaver's thorax, and CT volume—maps a trajectory drawn in VR into robot motion in the intervention room. In the VR scene, a CT-derived skin mesh is overlaid with a color map of the maximum Hounsfield unit along candidate needle paths, so the user immediately sees which insertions would hit bone and whether the robot's reach and joint limits make the trajectory feasible. The three planning scenarios differ only in how the needle path is specified; the underlying registration and feasibility checks are identical.

What would settle it

Compare the planning CT and a post-insertion CT using radio-opaque markers embedded in the phantom and at the target sites: if internal targets shift relative to the strapped phantom, the registered needle-tip error should decrease once that displacement is compensated. If it does not, the 5.30 mm off-axis error is genuine placement accuracy.

Watch

Extended reading notes

Core claim

The central claim is that a VR digital twin can support fully remote planning and execution of robotic post mortem biopsies with acceptable accuracy and usability. The paper demonstrates this by combining a CT-guided registration chain (robot base to tracking camera to a strapped phantom to the CT volume) with a VR application in which the user sees a life-size skin model, the robot, and CT slices in a shared coordinate frame. Users choose insertion points and needle angles in one of three interaction modes; the system immediately marks which trajectories are robot-feasible by overlaying a maximum-Hounsfield-unit color map on the skin. In the cadaver study, the mean off-axis error was 5.30 ±

Load-bearing premise

The accuracy numbers assume the cadaver's skin and internal organs stay fixed relative to the tracked phantom between the planning CT and each insertion; if the body shifts under the tension straps, the reported error could partly be target displacement rather than robot placement error.

Editorial extensions

If this is right

Remote planning and execution: a physician can run the whole biopsy workflow from a control room, eliminating direct contact with the cadaver and reducing infection risk.
No robotics expertise required: novice and expert users planned feasible trajectories after a five-minute introduction, and planning times dropped 33–53% across repeated insertions.
Systematic sampling becomes practical: because the robot places each needle consistently, the workflow suits large multi-target collections such as national autopsy registries.
Accuracy is comparable to prior robotic PM biopsy systems (target error 7.35 vs 7.19 mm), while surface puncture error is lower (2.62 mm).
Scalability to infectious disease outbreaks: the system is mobile and can be deployed where autopsies are otherwise avoided.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

Because the reported error is computed against the planning CT, part of the 5.30 mm off-axis value may reflect target displacement (e.g., tissue shifting after earlier insertions) rather than pure robot placement error; a marker-based repeat-CT experiment could separate the two.
The 65% histology confirmation rate suggests the limiting factor for small targets is not placement error alone but the biopsy needle and target size; a different needle design or intraprocedural confirmation could plausibly raise the hit rate.
The same digital-twin interaction could be adapted to living patients if respiratory and cardiac motion are added to the registration and planning, but that would require motion compensation the current post mortem setting does not need.
The strong preference for the 3D grab-and-orient modes over the conventional triplanar screen mode hints that pathologists may accept VR planning more readily if the interaction matches how they think about anatomy, not how legacy software displays slices.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Short take: this is a real systems paper, not a simulation stunt. The new bit is putting a VR digital twin in the control loop for fully remote post mortem biopsy, with three planning modes and evaluation on actual cadavers. That combination is genuinely new, and the authors document the workflow—hand-eye calibration, registration chain, latency, end-to-end tissue extraction—carefully enough that the system could be replicated. The usability data from 21 users is credible, and the learning-curve numbers go in the right direction. Credit also for reporting the 65% histopathology confirmation rate in the body and for explicitly disclosing the six excluded insertions.

The main soft spot is whether the headline accuracy number means what it says. Eq. (1) registration and the post-insertion CT error computation assume no tissue shift between planning and insertion. The paper itself says previous insertions can displace targets, and all 132 insertions were included in the error analysis, including later ones on the same target. So part of the reported 5.30 mm off-axis error may be target movement, not needle placement error. That is a legitimate confound, though not fatal: the surface-point error is lower and the histology success rate is consistent with small targets and shallow insertions. The stress-test note is directionally right.

The other issues are real but minor: six excluded needles (five loose due to low BMI, one touched by the EEF) could bias accuracy in either direction and should be reported both ways; the abstract says samples were successfully biopsied and histopathologically verified without giving the 65% rate, which overstates the result; and the 'alternative to conventional approaches' claim lacks a manual baseline, so it should be framed as feasibility.

I don't think the paper is circular or hiding anything. The comparison to their earlier robotic system [9] is a benchmark, not an input, and the registration chain is standard. The weakest assumption is standard for this kind of cadaver study, but the authors could address it by reporting per-target insertion order or excluding multiple insertions from the accuracy computation.

Who is this for? People working on remote robot control, forensic pathology, or VR surgical planning. It deserves a serious referee; the thesis is clear and the evaluation is honest enough that a revision can fix the flags. I'd engage with it—if I did forensic robotics, I'd cite it.

Referee Report

3 major / 5 minor

Summary. The paper presents a mobile robotic post-mortem biopsy system in which a physician plans needle trajectories in a virtual-reality digital twin and the robot executes the insertions in a physically separate intervention room. The system is evaluated in a usability study with 21 participants across three VR planning scenarios and in a clinical-feasibility study on three cadavers. The authors report 132 needle insertions with a mean off-axis error of 5.30±3.25 mm, a mean target error of 7.35±4.10 mm, and histopathological confirmation of 65% of tissue samples. The central claims are that fully remote VR planning is feasible and usable and that the system achieves clinically acceptable placement accuracy.

Significance. If the accuracy figure is valid, the system is a meaningful step toward reducing infection risk during post-mortem tissue sampling and enabling remote systematic biopsy collection. The paper provides a concrete and largely transparent evaluation protocol: an explicit registration chain (Eq. 1), three separate error measures, a multi-user usability study with physicians, and histopathological follow-up. The disclosure of excluded needles and the acknowledgment that targets can be displaced by prior insertions are also to the authors' credit. However, the central accuracy claim is threatened by the conflation of needle placement error with target motion, and the abstract overstates the histopathological result. The paper's contribution is promising, but the accuracy evidence needs strengthening before the headline number can be accepted at face value.

major comments (3)

[Section II.F / Section III.B, Eq. (1)] The headline off-axis error of 5.30±3.25 mm is computed for all 132 insertions, including the second and third insertion to the same target, by comparing the needle axis annotated in the post-insertion CT against the target position from the planning CT. The paper itself states in Section II.F that 'targets can be displaced due to previous needle insertions,' and indeed only tissue samples for the first insertion are extracted to avoid this bias. This creates a direct confound: for later insertions, the deviation between the annotated needle axis and the planned target may reflect target displacement rather than robot needle-placement error. Please report the error measures separately for first insertions only, or otherwise correct for target motion (e.g., by registering post-insertion CT to planning CT and updating target coordinates). Without this, the central accuracy claim is not sup
[Abstract / Section III.B] The abstract states 'Tissue samples were successfully biopsied and histopathologically verified,' which implies that the extracted samples were all confirmed. The results in Section III.B report histopathological confirmation in only 65% of samples, with the majority of misses for small structures such as the coronary artery. This is a substantial understatement of a limitation and should be corrected. Please report the 65% confirmation rate, or an equivalent qualification, in the abstract so that the summary matches the evidence.
[Section III.B (needle exclusion)] Five needles were removed from the study because they did not stay fixed after insertion, and one more was removed because it was touched by the robot EEF. Removing failure cases from the accuracy analysis can bias the reported error downward, especially because the five loose needles are attributed to the low-BMI cadaver (cadaver 2), making the exclusion potentially systematic rather than random. The disclosure is commendable, but the paper should quantify the sensitivity of the reported error to these exclusions, e.g., by reporting the error with the excluded needles included (with reasonable handling of their positions) or by reporting per-cadaver breakdowns that make the effect of the exclusion explicit.

minor comments (5)

[Discussion, comparison to [9]] The comparison of the target point error with Neidhardt et al. [9] (7.19±4.22 mm vs. 7.35±4.10 mm) is presented without any statistical test or confidence interval. Please clarify whether this is an informal comparison, and avoid drawing conclusions about equivalence without a quantitative basis.
[Fig. 6 caption] The caption appears truncated: 'In red the learning curve is indicated' and then the sentence about users being tasked is incomplete. Please rewrite for clarity.
[Section II.F] The text reads 'Gauge 13 biopsy needle'; standard notation is '13 Gauge' or '13G'. Also, please specify the needle diameter and length for reproducibility.
[Section II.B] The calibration time 358±32 s is reported, but it is not clear whether this includes the CT scan and table positioning or only the hand-eye calibration. Please clarify what the reported interval covers.
[Section II.C] The term 'digital twin' is used to describe the VR representation, but no formal definition is given. A brief statement of what aspects of the physical system are mirrored (robot kinematics, CT geometry, skin model) and which are not (e.g., tissue deformation, needle-tissue interaction) would help set expectations.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the central claims are empirical measurements of system accuracy and usability, not derivations from fitted inputs or self-cited premises.

full rationale

The paper is an empirical systems evaluation. The headline accuracy figures (target error 7.35±4.10 mm, off-axis error 5.30±3.25 mm, surface point error 2.62±1.76 mm) are measured quantities obtained by annotating needle axes in post-insertion CT and comparing them to planned target positions from the planning CT. This is an external measurement, not a prediction derived from a fitted parameter. Eq. (1), B_T_CT = B_T_TB · T_B_T_C · C_T_RM · R_M_T_SB · S_B_T_CT, is a calibration chain composed of independently estimated rigid transformations (hand-eye calibration, tracked marker positions, phantom geometry, CT segmentation); it is not constructed to reproduce the target error or any other reported outcome. The usability study compares three VR interaction methods with user ratings and planning times; these are direct empirical observations. The only self-citation to prior work [9] is used to compare a similar target error (7.19±4.22 mm vs. 7.35±4.10 mm) and to justify the phantom design as 'similar to [9]'; this comparison is not an input to the reported error and does not function as a load-bearing derivation. The limitation that targets may be displaced by previous insertions, and the corresponding decision to extract tissue only for first insertions, is a validity concern about whether the measured off-axis error conflates target displacement with placement error. That is a correctness/interpretation issue, not circularity: the error is still measured, not defined into existence by the paper's assumptions. Similarly, the 65% histopathological confirmation rate is an empirical outcome explicitly discussed with plausible causes (needle type, small targets, autolysis). No fitted input is renamed as a prediction, no uniqueness theorem is imported from the authors' prior work, and no ansatz is smuggled in via citation. The paper's central claims are therefore self-contained empirical evaluations rather than circular derivations.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

The central claim rests on standard rigid-registration assumptions and equipment choices rather than invented physical entities. The 15 mm depth offset is a hand-set compensation that affects reported errors, and the 10 mm feasibility spacing affects planning visualization. No new particles, forces, mediators, or conserved quantities are introduced.

free parameters (2)

Needle insertion depth offset = 15 mm subtracted from insertion depth
The system subtracts 15 mm from the insertion depth to position the biopsy punch at the target center, then virtually extends the needle by 15 mm during accuracy evaluation. This hand-set compensation directly enters the reported target and off-axis errors.
Feasibility sampling spacing = 10 mm
Feasible robot insertion paths are checked at discrete colormap points separated by 10 mm. This spacing was chosen to balance inference time (18.80 s) against spatial resolution and affects which paths the user sees as feasible during planning.

assumptions (3)

domain assumption Cadavers are static during the procedure, with no respiratory or cardiac motion.
The introduction states that vital structures do not continuously translate in the absence of respiratory motion, which is what makes unattended robotic insertion tractable. This is reasonable for post mortem tissue but is an assumption about the clinical setting.
domain assumption The rigid transforms of Eq. (1) remain valid between registration, planning, and insertion.
The registration chain from robot base to CT assumes the phantom, cadaver, CT table, and robot base do not move after calibration. Any shift in these components would enter all reported error metrics.
domain assumption The annotated needle in the post-insertion CT represents the placed needle, and the planned target remains at its pre-insertion CT location.
Accuracy is computed against a target defined in the planning CT, with no independent measurement of target displacement caused by needle insertion or prior insertions. The authors acknowledge this risk in Section II.F when they avoid re-biopsying the same target.

how reviews work

0 comments

Cite this review

Pith. "Pith review of A Digital Twin for Robotic Post Mortem Tissue Sampling using Virtual Reality." pith.science (2026). https://pith.science/paper/7TBP2QSS

@misc{pith2026250902760,
  author       = {Pith},
  title        = {Pith review of: A Digital Twin for Robotic Post Mortem Tissue Sampling using Virtual Reality},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7TBP2QSS}},
  note         = {Machine review of arXiv:2509.02760}
}

read the original abstract

Studying tissue samples obtained during autopsies is the gold standard when diagnosing the cause of death and for understanding disease pathophysiology. Recently, the interest in post mortem minimally invasive biopsies has grown which is a less destructive approach in comparison to an open autopsy and reduces the risk of infection. While manual biopsies under ultrasound guidance are more widely performed, robotic post mortem biopsies have been recently proposed. This approach can further reduce the risk of infection for physicians. However, planning of the procedure and control of the robot need to be efficient and usable. We explore a virtual reality setup with a digital twin to realize fully remote planning and control of robotic post mortem biopsies. The setup is evaluated with forensic pathologists in a usability study for three interaction methods. Furthermore, we evaluate clinical feasibility and evaluate the system with three human cadavers. Overall, 132 needle insertions were performed with an off-axis needle placement error of 5.30+-3.25 mm. Tissue samples were successfully biopsied and histopathologically verified. Users reported a very intuitive needle placement approach, indicating that the system is a promising, precise, and low-risk alternative to conventional approaches.

Figures

Figures reproduced from arXiv: 2509.02760 by the authors.

**Figure 1.** Virtual reality guided tissue biopsy with a robot: Top: The user defines biopsy targets in CT images and plans 3-dimensional insertion paths inside a custom-designed VR application. The system allows fully remote planning and control of robotic biopsies. Bottom: An LBR Med robot (A) mounted to a mobile platform (B) positions needles inside a cadaver (C). Before insertions, a CT scan (D) is acquired. The robot is cal… view at source ↗

**Figure 2.** System communication: Overview of the communication links between the control and the intervention room. BT CT CTT SB SBT RM RMT C CT TB TBT B BT EEF EEFT N (a) Intervention Room WT B BT CT BT EEF EEFT N WT Q3 (b) VR Application [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Relevant transformations: System transforms applied in the intervention room (a) and in the VR application (b). Transforms indicated in red are identical for both. of respiratory motion. To this end, a fixed ceiling-mounted robotic system for inserting biopsy needles has been presented [8]. More recently, a flexible approach using a lightweight robotic arm has been demonstrated [9]. This system can be fully deploye… view at source ↗

Figures from the paper (5 more)

**Figure 4.** Figure 4: VR needle path planning scenarios: We design 3 scenarios to plan needle insertion points. Each scenario contains a CT scanner, a screen, and robot. Projected on the skin model is the maximum Hounsfield unit between insertion and target point. Saturated colors indicate …

**Figure 2.** Figure 2: We use a PACS server to transfer CT image data to the [PITH_FULL_IMAGE:figures/full_fig_p003_2.png]

**Figure 5.** Figure 5: Clinical Workflow: Individual steps during VR assisted needle placement. and the VR workstation using the Meta Quest Link-App (ver. 72.0.0.500.353). This communication protocol allows us to perform high-resolution renderings on the GPU (RTX 4090, Nvidia, California, US…

**Figure 6.** Figure 6: Results usability study: (a) Planning time for all insertions performed. (b-d) Planning time for subsequent insertions. In red the learning curve is indicated. (e) The users (n=21) were tasked (e) to rate the individual planning scenarios 1-3 and (f) to rate the overal…

**Figure 7.** Figure 7: System Accuracy: (a) Error of all needles inserted, (b) target error for individual organs. previous needle insertions. In total, 46 tissue probes were extracted. After needle placement, the CT imaging protocol was repeated. For each inserted needle, we manually annota…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

16 extracted references · 16 canonical work pages

[9]

Robotic tissue sampling for safe post-mortem biopsy in infectious corpses,

M. Neidhardt et al., “Robotic tissue sampling for safe post-mortem biopsy in infectious corpses,” IEEE T-MRB, vol. 4, pp. 94–105, 2022

work page 2022
[1]

Contamination of personal protective equipment during covid-19 autopsies,

J. M. Brandner et al., “Contamination of personal protective equipment during covid-19 autopsies,” Virchows Archiv, vol. 480, no. 3, pp. 519– 528, 2022

work page 2022
[2]

Postmortem minimally invasive autopsy in critically ill covid-19 patients at the bedside: A proof-of-concept study at the icu,

T. Lahmer et al., “Postmortem minimally invasive autopsy in critically ill covid-19 patients at the bedside: A proof-of-concept study at the icu,” Diagnostics, vol. 14, no. 3, p. 294, 2024

work page 2024
[3]

Minimally invasive tissue sampling via post mortem ultrasound: A feasible tool (not only) in infectious diseases—a case report,

A. Terence Azeke et al., “Minimally invasive tissue sampling via post mortem ultrasound: A feasible tool (not only) in infectious diseases—a case report,” Diagnostics, vol. 13, no. 16, p. 2643, 2023

work page 2023
[4]

Clinical evaluation of a robotic system for precise ct- guided percutaneous procedures,

S. Levy et al., “Clinical evaluation of a robotic system for precise ct- guided percutaneous procedures,” Abdominal radiology, vol. 46, no. 10, pp. 5007–5016, 2021

work page 2021
[5]

Percutaneous liver interventions with robotic systems: a systematic review of available clinical solutions,

S. Bodard et al., “Percutaneous liver interventions with robotic systems: a systematic review of available clinical solutions,” Brit. J. Radiol., vol. 96, no. 1152, p. 20230620, 2023

work page 2023
[6]

C. Moser et al., “A novel laser navigation system reduces radiation exposure and improves accuracy and workflow of ct-guided spinal interventions: a prospective, randomized, controlled, clinical trial in comparison to conventional freehand puncture,” European Journal of Radiology, vol. 82, no. 4, pp. 627–632, 2013

work page 2013
[7]

Collaborative robotic biopsy with trajectory guidance and needle tip force feedback,

R. Mieling et al., “Collaborative robotic biopsy with trajectory guidance and needle tip force feedback,” in 2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023, pp. 6893–6900

work page 2023

Show all 16 references

[8]

Virtobot–a multi-functional robotic system for 3d surface scanning and automatic post mortem biopsy,

L. C. Ebert et al., “Virtobot–a multi-functional robotic system for 3d surface scanning and automatic post mortem biopsy,” Int J Med Robot Comput Assist Surg, vol. 6, no. 1, pp. 18–27, 2010

2010
[10]

Augmented reality for surgical navigation: A review of advanced needle guidance systems for percutaneous tumor ablation,

M. Evans et al., “Augmented reality for surgical navigation: A review of advanced needle guidance systems for percutaneous tumor ablation,” Radiology: Imaging Cancer, vol. 7, no. 1, p. e230154, 2025

2025
[11]

Diagnostic performance of a next-generation virtual/augmented reality headset: A pilot study of diverticulitis on ct,

P. M. Murphy et al., “Diagnostic performance of a next-generation virtual/augmented reality headset: A pilot study of diverticulitis on ct,” JIIM, pp. 1–8, 2024

2024
[12]

Slicervr for medical intervention training and planning in immersive virtual reality,

C. Pinter et al., “Slicervr for medical intervention training and planning in immersive virtual reality,” IEEE T-MRB, vol. 2, no. 2, pp. 108–117, 2020

2020
[13]

Non-orthogonal tool/flange and robot/world calibration,

F. Ernst et al., “Non-orthogonal tool/flange and robot/world calibration,” Int J Med Robot Comput Assist Surg, vol. 8, no. 4, pp. 407–420, 2012

2012
[14]

I. A. Sucan and S. Chitta, “Moveit,” 2012. [Online]. Available: moveit.ros.org

2012
[15]

Assessment of the dimensions of coronary arteries for the manifestation of coronary artery disease,

M. Muneeb et al., “Assessment of the dimensions of coronary arteries for the manifestation of coronary artery disease,” Cureus, vol. 15, 2023

2023
[16]

First report from the german covid-19 autopsy registry,

S. von Stillfried et al., “First report from the german covid-19 autopsy registry,” The Lancet Regional Health–Europe, vol. 15, 2022

2022

Pith tools

Reviewed August 5, 2026 · model on record in the stance chip above.

[1] [9]

Robotic tissue sampling for safe post-mortem biopsy in infectious corpses,

M. Neidhardt et al., “Robotic tissue sampling for safe post-mortem biopsy in infectious corpses,” IEEE T-MRB, vol. 4, pp. 94–105, 2022

work page 2022

[2] [1]

Contamination of personal protective equipment during covid-19 autopsies,

J. M. Brandner et al., “Contamination of personal protective equipment during covid-19 autopsies,” Virchows Archiv, vol. 480, no. 3, pp. 519– 528, 2022

work page 2022

[3] [2]

Postmortem minimally invasive autopsy in critically ill covid-19 patients at the bedside: A proof-of-concept study at the icu,

T. Lahmer et al., “Postmortem minimally invasive autopsy in critically ill covid-19 patients at the bedside: A proof-of-concept study at the icu,” Diagnostics, vol. 14, no. 3, p. 294, 2024

work page 2024

[4] [3]

Minimally invasive tissue sampling via post mortem ultrasound: A feasible tool (not only) in infectious diseases—a case report,

A. Terence Azeke et al., “Minimally invasive tissue sampling via post mortem ultrasound: A feasible tool (not only) in infectious diseases—a case report,” Diagnostics, vol. 13, no. 16, p. 2643, 2023

work page 2023

[5] [4]

Clinical evaluation of a robotic system for precise ct- guided percutaneous procedures,

S. Levy et al., “Clinical evaluation of a robotic system for precise ct- guided percutaneous procedures,” Abdominal radiology, vol. 46, no. 10, pp. 5007–5016, 2021

work page 2021

[6] [5]

Percutaneous liver interventions with robotic systems: a systematic review of available clinical solutions,

S. Bodard et al., “Percutaneous liver interventions with robotic systems: a systematic review of available clinical solutions,” Brit. J. Radiol., vol. 96, no. 1152, p. 20230620, 2023

work page 2023

[7] [6]

C. Moser et al., “A novel laser navigation system reduces radiation exposure and improves accuracy and workflow of ct-guided spinal interventions: a prospective, randomized, controlled, clinical trial in comparison to conventional freehand puncture,” European Journal of Radiology, vol. 82, no. 4, pp. 627–632, 2013

work page 2013

[8] [7]

Collaborative robotic biopsy with trajectory guidance and needle tip force feedback,

R. Mieling et al., “Collaborative robotic biopsy with trajectory guidance and needle tip force feedback,” in 2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2023, pp. 6893–6900

work page 2023