Wound3DAssist: A Practical Framework for 3D Wound Assessment

Chuong Nguyen; Clinton Fookes; David Ahmedt-Aristizabal; Jeremy Oorloff; L\'eo Lebrat; Mohammad Ali Armin; Olivier Salvado; Remi Chierchia; Rodrigo Santa Cruz; Yulia Arzhaeva

REVIEW 3 major objections 5 minor 60 references

Wound3DAssist: A Practical Framework for 3D Wound Assessment

T0 review · 3 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A short smartphone video can yield a 3D wound model with automated depth, area, and tissue metrics.

desk verdict A transparent, well-engineered system paper whose real novelty is the surface-cover depth estimation, but the headline millimeter-level accuracy claim is not yet supported for depth because that component has no ground-truth validation. read the letter →

arxiv 2508.17635 v1 pith:S3UBQWC7 submitted 2025-08-25 cs.CV

Remi Chierchia , Rodrigo Santa Cruz , Léo Lebrat , Yulia Arzhaeva , Mohammad Ali Armin , Jeremy Oorloff , Chuong Nguyen , Olivier Salvado

show 2 more authors

Clinton Fookes David Ahmedt-Aristizabal

This is my paper · ORCID

classification cs.CV

keywords 3Dwoundassessmentmonocularphotogrammetrysegmentationtissueclassificationdepthmeasurementsmartphonevideolongitudinalmonitoringmeshanalysis

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

This paper sets out to show that accurate 3D wound assessment no longer requires specialized scanners: a short handheld smartphone video, processed by photogrammetry and neural segmentation, can be turned into a colored 3D model from which area, perimeter, length, width, depth, and tissue composition are computed automatically. The motivation is that current manual methods, such as ruler and tracing techniques, are subjective, view-dependent, and cannot capture wound depth on curved anatomy. The authors validate the claim on synthetic models with known geometry, silicone phantoms, and real patient videos, reporting millimeter-level reconstruction accuracy and statistically improved 3D segmentation over single-view 2D segmentation for wound bed and periwound regions. Full documentation runs in under 20 minutes, which the paper argues makes the workflow feasible for routine clinical use.

What carries the argument

The load-bearing object is the wound bed surface cover: a virtual smooth surface fitted to the wound perimeter vertices using radial basis function interpolation with a thin-plate spline kernel. This surface stands in for where the skin was before injury, and every depth and volume-related measurement is computed as the vertical distance from a wound-bed vertex to this cover. It also carries the geodesic length and width calculations, since those paths are projected onto the cover before their arc lengths are measured.

What would settle it

On a curved calibration phantom with a known cavity geometry, record a handheld video, run the full pipeline, and compare the computed depth to caliper-measured depth; if depth error grows systematically with surface curvature, the smooth cover does not represent pre-injury skin.

Watch

Extended reading notes

Core claim

On its own terms, the paper argues that monocular videogrammetry of a wound is sufficient: after key-frame selection, an open-source structure-from-motion pipeline builds a textured 3D mesh; a fine-tuned transformer-based segmentation network labels wound bed, periwound, and tissue classes in 2D; majority voting over visible views, weighted by viewing angle, projects those labels onto the mesh; and all clinical measurements are then made directly on the mesh. Depth is defined relative to a smooth surface cover fitted to the wound perimeter, and scale is recovered from fiducial markers placed in the scene. The evaluation claims sub-millimeter precision on phantoms, millimeter-level surface agreement on real wounds, view-consistent segmentation that is at least as good as single-view 2D segmentation, and longitudinal trends that match both healing and non-healing courses.

Load-bearing premise

The depth and volume numbers all rest on the assumption that the smooth surface fitted to the wound perimeter is where the skin was before injury; if that cover is wrong on curved anatomy, every depth metric is wrong regardless of reconstruction quality.

Editorial extensions

If this is right

Wound area, perimeter, length, width, and depth can be measured without touching the wound, from a video captured on an ordinary phone, making repeated assessment practical in telehealth and home care.
Because measurements are taken on the reconstructed 3D surface rather than a single photograph, they should be view-independent and robust to camera motion during capture.
The same reconstructed mesh can be revisited across consultations; the paper shows area shrinkage and a slough-to-granulation transition in a healing patient and area growth in a non-healing patient.
View-weighted 2D-to-3D label voting yields wound bed and periwound segmentation that is at least as accurate as single-view 2D segmentation, so the 3D step does not sacrifice tissue classification quality.
The modular design lets reconstruction or segmentation components be swapped without changing the measurement layer, so future accuracy improvements can enter the pipeline incrementally.

Reading between the lines

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

If the perimeter-cover assumption holds on curved anatomy, the same measurement layer could be applied to other cavity-like surface conditions, such as pressure injuries or surgical wounds, without retraining the geometry pipeline.
The view-weighted voting rule predicts a testable property: deliberately adding oblique or blurry frames should barely change the final 3D labels, because low-confidence views are down-weighted; a perturbation study varying acquisition angle would quantify this robustness.
A natural stress test the paper does not run is scan-rescan repeatability: recording the same wound twice after repositioning the camera should yield nearly identical depth and area, which would separate reconstruction noise from true healing change.
As neural rendering methods mature, the same cover-and-voting machinery could operate on their meshes, making the measurement step independent of whichever reconstruction backend proves fastest or most accurate.

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

You should know two things about this one. First, it is a competent systems-integration paper: smartphone video in, a textured 3D mesh out, with 2D-to-3D segmentation and a set of clinically motivated wound metrics. The genuinely new piece is the normal-weighted 2D-to-3D majority voting and the TPS surface-cover used to estimate depth, area, and volume. That combination is not in the cited prior work. Second, the paper's headline claim, "millimeter-level accuracy", overreaches for the depth metric specifically, because depth is never checked against any ground truth. The stress-test note is right: the surface cover is fitted to perimeter vertices with a bending-energy-minimizing TPS, and all depth/volume numbers are distances to that virtual surface. On curved anatomy that surface can deviate a lot from the true pre-injury skin, and no experiment quantifies that error. The clinical comparison uses Wintape and ruler, both of which ignore depth. So the abstract's accuracy claim is really only supported for reconstruction geometry, not for the depth measurement that justifies 3D over 2D.

Credit where it is due: the authors are unusually candid about limitations. They explicitly say the ArUco alignment failed on clinical data, that manual alignment was used instead, that the clinical set is only four cases, and that necrotic and epithelial classes are omitted due to sparse data. The modular design and the decision to stick with Meshroom after comparing against NeRF and Gaussian Splatting is reasonable. The longitudinal case studies, while qualitative, are a useful sanity check. The 18-minute runtime is a practical claim that is backed by measurements.

Soft spots, in proportion. The four-case clinical evaluation is small, and the approximate 1 mm AD is reported without error bars or per-case detail. Two of the three validation tiers (digital and silicone) are explicitly deferred to the authors' own prior papers [23] and [24]; as a reviewer you would want those numbers reproduced or at least summarized. The statistical significance claim for the 2D-to-3D segmentation improvement is based on a Wilcoxon test, but the text says it only holds for wound bed and periwound, not for granulation and slough, so the claim that 2D-to-3D outperforms 2D is weaker than it looks. None of these are fatal, but they add up.

Who this is for: a clinical imaging researcher or a referee for a medical-vision venue. It is a useful integration of existing building blocks and it provides a reasonable template for validation on real patient data. It deserves peer review, but the depth and volume claims need either a phantom with known ground-truth depth or a validation study against a structured-light scanner or CT, plus more than four clinical cases, before the abstract's headline can stand. I would send it out, with a request for major revisions.

Referee Report

3 major / 5 minor

Summary. The paper presents Wound3DAssist, a modular framework for 3D wound assessment from monocular smartphone videos. The pipeline combines photogrammetric reconstruction (Meshroom), a fine-tuned SegFormer 2D segmentation network whose predictions are rasterized onto the 3D mesh, and a geometry-processing module that computes wound bed perimeter, length, width, surface area, depth, and tissue composition. The authors evaluate the framework across three tiers: a synthetic digital wound dataset with known geometry, silicone phantoms, and a prospective clinical dataset from nine patients (with four reconstructions). They report sub-millimeter precision on the phantom dataset (deferred to prior work [24]), an average point-wise reconstruction error on the four clinical cases, improved 2D-to-3D segmentation relative to 2D baselines with statistical significance for wound bed and periwound, and longitudinal tracking of wound progression on two patients. The full documentation pipeline is reported to run in approximately 18 minutes on a high-end workstation.

Significance. If the depth and volume measurements were properly validated, the framework would be a practically relevant contribution: it uses consumer-grade hardware, produces colored 3D models, integrates segmentation and tissue classification, and demonstrates longitudinal tracking. The modular design and the systematic comparison of 2D versus 2D-to-3D segmentation mapping are useful methodological contributions. However, the principal claim that distinguishes this work from 2D assessment—millimeter-accurate depth measurement—is not supported by the evidence in the manuscript. The digital and silicone validation tiers are deferred to the authors' prior papers [23], [24], and the clinical evaluation has no depth ground truth, relies on manual alignment, and uses only four cases. The paper is best characterized as a promising systems paper whose central quantitative claim requires additional experiments.

major comments (3)

[Sec. IV-C4 and Eq. (3)] The wound bed depth metric is defined as the vertical distance between wound bed vertices and the TPS/RBF surface cover fitted to the perimeter vertices. This surface is described as 'intended to represent the skin's original location prior to injury,' but no experiment validates that the fitted surface matches the pre-injury skin geometry. The digital dataset with known geometry (Sec. II-A) is used only for reconstruction evaluation (Sec. V-A1), and the clinical comparison (Sec. V-C1) uses ruler and Wintape, neither of which measures depth. Consequently, the abstract's claim of 'millimeter-level accuracy' is not supported for the depth measurement, which is the principal justification for 3D assessment over 2D. Please add a validation experiment, for example by computing depth errors against the known geometry in the digital dataset or by comparing the TPS cover to a ground-truth skin surface on a phantom.
[Sec. V-A3] The clinical reconstruction evaluation is limited to four cases and requires manual alignment of the reconstruction to the Revopoint reference because the ArUco markers were unreliable in the RGB stream. The manuscript does not state the numeric value of the average point-wise error (the symbol is missing), nor does it quantify the uncertainty introduced by manual alignment. Moreover, the reported errors are concentrated at the wound bed boundaries (Fig. 7), which is exactly the region used for perimeter and depth measurement. Please report per-case errors with the exact average, and provide a sensitivity analysis of the manual alignment, or downgrade the clinical accuracy claim accordingly.
[Sec. V-B] The segmentation evaluation is performed on 2D re-projections of the 3D segmentation against 2D expert annotations; there is no ground-truth 3D segmentation. The Wilcoxon signed-rank test yields statistical significance for wound bed and periwound, but not for granulation or slough, which are the two tissue classes with sufficient samples. The claim that the 2D-to-3D method improves tissue segmentation is therefore only partially supported. Please provide additional evidence for the tissue classes, or refine the claim to reflect the classes for which significance was shown.

minor comments (5)

[Throughout] The manuscript contains many garbled characters and placeholder values (e.g., Sec. V-A3 'approximately � mm', Sec. IV-A1 '� � �� frames', and the figure captions for Figs. 5, 10, 11, 12, and 13). These must be corrected to allow proper review and to make the reported results readable.
[Sec. IV-C1, Eq. (5)] The B-spline perimeter estimate introduces a smoothness regularization parameter 's', but the manuscript does not report the value used in the experiments or any sensitivity analysis for this parameter.
[Sec. IV-C2, Eq. (7)] The geodesic distance is computed by projecting intermediate Euclidean points onto the surface cover and fitting a B-spline; please clarify how the projection is well-defined when the line segment between perimeter vertices lies partially outside the convex hull of the fitted surface domain.
[Sec. V-C1] The statement 'Similar outcomes were observed across additional cases' is not backed by a quantitative comparison. Please include a table reporting framework and manual measurements (area, perimeter, and depth, where available) for all four prospective cases.
[Sec. IV-B2, Eq. (2)] The weighting factor in Eq. (1) and the voting scheme in Eq. (2) are described with notation that is partially illegible in the submitted version; please ensure all symbols and index ranges are typeset correctly.

Circularity Check

0 steps flagged · score 0.0 of 10

No definitional circularity: all wound metrics are computed by explicit formulas from the reconstructed mesh, and the accuracy claims are anchored to external benchmarks (prior public datasets and Revopoint clinical scans) rather than to the framework's own fitted outputs.

full rationale

The derivation chain in Wound3DAssist is a modular engineering pipeline rather than a predictive derivation. The 3D reconstruction is checked against ground truth from the digital dataset, silicone phantoms, and Revopoint scans of real patients. Although the detailed digital and silicone numbers are deferred to the authors' earlier papers [23] and [24], those are public benchmark datasets with own acquisition protocols and quantitative results; citing them is self-citation but not circularity, because the cited benchmarks are externally falsifiable references rather than the current paper's fitted outputs. The segmentation evaluation compares 2D re-projections of 3D labels to expert 2D annotations via DSC, not merely to the same 2D predictions used as input. The wound measurements (area, perimeter, length, width, depth, tissue composition) are computed from explicit formulas (Eqs. 3-7 and Section IV-C) operating on the segmented mesh. The only step that could look circular is the TPS/RBF surface cover: it is fitted to wound-perimeter vertices and then used to define depth as vertical distance to that surface. This is an operational definition of depth, not a fitted parameter renamed as a prediction: the surface cover is not fitted to depth values, and the depth values do not feed back into the surface fit. Whether the fitted surface truly represents the pre-injury skin contour is an unvalidated clinical assumption and a genuine correctness risk on curved anatomy, but it is not a circularity by construction. No equation in the paper reduces to its own input, and no central claim is justified solely by an overlapping-author citation.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The framework does not derive constants from first principles; it relies on standard photogrammetry and segmentation machinery plus two load-bearing modeling choices (the TPS surface cover as the pre-injury skin reference, and manual alignment for clinical benchmarking). The hyperparameters are hand-set rather than fitted to data, so the ledger is short.

free parameters (4)

Number of selected keyframes = not reported (text shows a blank where a number should be)
The number of frames used for reconstruction is chosen by hand as a balance between coverage and quality; it directly affects reconstruction detail and runtime.
Maximum SIFT features per image = 50,000
Custom Meshroom setting to increase detail, with a minor processing time increase (Sec. IV-A2).
Obliqueness threshold cos theta = 0.5
Views where the face normal and camera ray have cos theta below 0.5 are excluded from label voting (Sec. IV-B2).
B-spline perimeter smoothness regularization lambda = not fixed; should be adapted to wound type
Regularization in Eq. 5; the paper notes it should be tuned per wound type, so it is not a universal constant.

assumptions (5)

domain assumption The TPS surface cover fitted to the wound perimeter represents the original skin surface before injury.
Depth and volume are computed relative to this surface; if the fit is wrong on curved anatomies, the metrics are wrong. Eq. 3-4, Sec. IV-C.
domain assumption The 2D segmentation model generalizes to prospective clinical video frames from different devices and settings.
The segmentation network was trained on retrospective images and applied to new video frames; performance is validated only on a small subset. Sec. II-C, V.B.
domain assumption ArUco markers provide accurate metric scale via DLT triangulation when captured reliably.
The scale factor converts all measurements to real-world units; in the clinical dataset markers were not reliably captured, forcing manual alignment. Sec. IV-C6, V.A.3.
domain assumption Meshroom photogrammetry produces sufficiently accurate geometry for wound scenes.
The framework relies on the open-source pipeline's output; accuracy depends on texture, lighting, and motion blur as discussed in Sec. V.A.
domain assumption Manual alignment of reconstructions to Revopoint point clouds is a valid evaluation reference.
Clinical error estimates are obtained after manual alignment, which the authors acknowledge introduces potential errors. Sec. V.A.3.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Wound3DAssist: A Practical Framework for 3D Wound Assessment." pith.science (2026). https://pith.science/paper/S3UBQWC7

@misc{pith2026250817635,
  author       = {Pith},
  title        = {Pith review of: Wound3DAssist: A Practical Framework for 3D Wound Assessment},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/S3UBQWC7}},
  note         = {Machine review of arXiv:2508.17635}
}

read the original abstract

Managing chronic wounds remains a major healthcare challenge, with clinical assessment often relying on subjective and time-consuming manual documentation methods. Although 2D digital videometry frameworks aided the measurement process, these approaches struggle with perspective distortion, a limited field of view, and an inability to capture wound depth, especially in anatomically complex or curved regions. To overcome these limitations, we present Wound3DAssist, a practical framework for 3D wound assessment using monocular consumer-grade videos. Our framework generates accurate 3D models from short handheld smartphone video recordings, enabling non-contact, automatic measurements that are view-independent and robust to camera motion. We integrate 3D reconstruction, wound segmentation, tissue classification, and periwound analysis into a modular workflow. We evaluate Wound3DAssist across digital models with known geometry, silicone phantoms, and real patients. Results show that the framework supports high-quality wound bed visualization, millimeter-level accuracy, and reliable tissue composition analysis. Full assessments are completed in under 20 minutes, demonstrating feasibility for real-world clinical use.

Review history (2 revisions) →

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 58 canonical work pages

[23]

Wound measurement by RGB-D camera,

D. Filko, R. Cupec, and E. K. Nyarko, “Wound measurement by RGB-D camera,” Mach. Vis. Appl. , vol. 29, pp. 633–654, 2018

work page 2018
[24]

Three-dimensional wound reconstruction using point descriptors: A comparative study,

J. R. Souto, F. M. Barbosa, and B. M. Carvalho, “Three-dimensional wound reconstruction using point descriptors: A comparative study,” in Proc. IEEE Int. Symp. Comput.-Based Med. Syst. (CBMS) , 2023, pp. 41–46

work page 2023
[1]

It was curated and used to train and validate our 2D segmentation model, a core component of the 3D segmentation pipeline

Retrospective data: This dataset includes 2D wound images captured in simulated clinical settings and shared under a data use agreement. It was curated and used to train and validate our 2D segmentation model, a core component of the 3D segmentation pipeline. The images were not used for 3D reconstruction or measurement tasks

work page
[2]

Videos were recorded using iPhones and Logitech cameras, while a Revopoint 3D scanner provided geometric ground truth

Prospective data: This dataset is collected in collabora- tion with clinical researchers, following a structured acquisition protocol to ensure consistency and suitability for 3D recon- struction, to enable real-world validation. Videos were recorded using iPhones and Logitech cameras, while a Revopoint 3D scanner provided geometric ground truth. The whol...

work page
[3]

By directly operating on the mesh structure and leveraging surface triangulation, this method provides precise and robust estimates

Wound bed and periwound surface area: The wound bed and periwound areas are computed by summing the areas of triangular mesh faces assigned to each respective segmentation label in the reconstructed surface. By directly operating on the mesh structure and leveraging surface triangulation, this method provides precise and robust estimates

work page
[4]

These frames are critical for accurately reconstructing the wound’s shape and surrounding tissue

Pre-processing: To create a high-quality 3D wound model, our framework selects a set of sharp frames that capture the wound from multiple angles. These frames are critical for accurately reconstructing the wound’s shape and surrounding tissue. Typically, � � �� frames are used, which provides a good balance between detail coverage and visual quality. Whil...

work page
[5]

3D reconstruction methods: Currently, our framework adopts the Meshroom pipeline [32] for 3D reconstruction. Specifically, we applied two custom settings to optimize for our dataset: i) Single-Camera Model : Tailored for monocular video input; ii) High-Density SIFT Features: Up to 50,000 SIFT features per image to improve detail, with a minor increase in ...

work page
[6]

2D segmentation: Our method builds upon the SegFormer architecture [30], specifically the MiT-b5 variant, which we fine- tuned to predict 2D segmentations of the wound bed, periwound region, and four tissue classes—granulation, necrotic, slough, and epithelial. Specifically, we fine-tuned the model for three distinct tasks: binary segmentation of the woun...

work page

Show all 60 references

[7]

For each available 2D view, we determine vertex visibility and assign a label by aggregating predictions using a majority voting scheme

2D-to-3D segmentation mapping: To obtain a 3D seg- mentation, we employ a rasterization algorithm [33] to project the 3D reconstruction mesh onto the predicted 2D segmentation masks. For each available 2D view, we determine vertex visibility and assign a label by aggregating p...
[8]

Wound boundaries are extracted by identifying mesh edges incident to a single triangle; the vertices of these edges define the wound perimeter

Wound bed perimeter: To compute the perimeter of the wound bed, we first isolate individual wounds by extracting the connected components from the segmented mesh. Wound boundaries are extracted by identifying mesh edges incident to a single triangle; the vertices of these edge...
[9]

Let � � �� be the set of wound bed perimeter vertices

Wound bed length and width: The length and width represent the two largest dimensions of the wound bed. Let � � �� be the set of wound bed perimeter vertices. The wound length is defined as the maximum distance between any pair of perimeter vertices: �� s...
[10]

Wound 3D geometrical feature estimation using poisson reconstruction,

J. M. Juszczyk, A. Wijata, J. Czajkowska, M. Krecichwost, M. Rudzki, M. Biesok, B. Pyci ´nski, J. Majewski, J. Kostecki, and E. Pietka, “Wound 3D geometrical feature estimation using poisson reconstruction,” IEEE Access, vol. 9, pp. 7894–7907, 2020

2020
[11]

A positive depth indicates a protrusion of the wound bed above the surface cover, while a negative depth corresponds to a depression below the surface

Wound bed depth: Given the reoriented wound bed and its corresponding surface cover, the depth at a vertex is defined as the vertical distance along the �-axis between the vertex and its projection onto the surface cover. A positive depth indicates a protrusion of the wound be...
[12]

The percentage of coverage is then 6 the ratio to the total wound bed area

Wound bed tissue composition: To analyze the tissue composition of the wound bed, we compute the surface area covered by each tissue type. The percentage of coverage is then 6 the ratio to the total wound bed area. This approach provides a quantitative assessment of wound bed ...
[13]

Annotations were provided for the periwound region, wound bed, and four tissue types: granulation (23 images), slough (37 images), necrotic (3 images), and epithelial (2 images)

[25] [26] [27] [5] [6] Ours Recent photogrammetry method � � � Recent segmentation method � � � Automatic wound segmentation � � � Wound multi-class tissue segmentation � � � � Wound depth/volume assessment � � � data contains images from 9 patients and includes wounds of dive...
[14]

We use ArUco markers placed near the wound and captured in the video sequence

Scale retrieval: A well-known limitation of 3D recon- struction from images is that the resulting model is defined only up to a similarity transformation relative to the real-world object [36]. We use ArUco markers placed near the wound and captured in the video sequence. The ...
[15]

Higher-resolution images consistently improved reconstruction quality, while increasing frame count offered diminishing returns

Results on digital wounds dataset: We evaluated re- construction accuracy using the digital wound dataset (Sec- tion II-A), which provides ground-truth geometry for objective assessment. Higher-resolution images consistently improved reconstruction quality, while increasing fr...
[16]

Recent neural rendering-based methods have demonstrated promising results for reconstruction accuracy, often superior to traditional pipelines such as Meshroom

Results on silicone wounds dataset: On the silicone dataset (Section II-B), we evaluated the accuracy and preci- sion of multiple 3D reconstruction methods across different recording devices. Recent neural rendering-based methods have demonstrated promising results for reconst...
[17]

For evaluation purposes, we focused on a subset of cases for which both iPhone video recordings and 3D scans from a Revopoint scanner were available

Results on clinical wounds dataset: We performed a quan- titative analysis using wound recordings collected from actual patients in healthcare settings (as described in Section II-C). For evaluation purposes, we focused on a subset of cases for which both iPhone video recordin...
[18]

11 illustrates a comparison between manual clinician-reported measurements and framework-generated metrics for a represen- tative case

Comparison of wound documentation methods: Fig. 11 illustrates a comparison between manual clinician-reported measurements and framework-generated metrics for a represen- tative case. Manual methods, especially ruler-based estimation, can show variability across sessions, part...
[19]

Longitudinal tracking of wound documentation: We also explored the use of our framework for tracking wound progression across multiple visits in a subset of patients. Fig. 12 presents results from �� . Over approximately three months, the wound showed clear signs of healing: t...

2022
[20]

An economic evaluation of the impact, cost, and medicare policy implications of chronic nonhealing wounds,

S. R. Nussbaum, M. J. Carter, C. E. Fife, J. DaVanzo, R. Haught, M. Nusgart, and D. Cartwright, “An economic evaluation of the impact, cost, and medicare policy implications of chronic nonhealing wounds,” V alue Health, vol. 21, no. 1, pp. 27–32, 2018

2018
[21]

Human wound and its burden: updated 2020 compendium of estimates,

C. K. Sen, “Human wound and its burden: updated 2020 compendium of estimates,” Adv. Wound Care, vol. 10, no. 5, pp. 281–292, 2021

2020
[22]

A comparison of wound area measurement techniques: visitrak versus photography,

A. C. Chang, B. Dearman, and J. E. Greenwood, “A comparison of wound area measurement techniques: visitrak versus photography,” Eplasty, vol. 11, p. e18, 2011

2011
[25]

Multi-view data augmentation to improve wound segmentation on 3D surface model by deep learning,

R. Niri, E. Gutierrez, H. Douzi, Y . Lucas, S. Treuillet, B. Castañeda, and I. Hernandez, “Multi-view data augmentation to improve wound segmentation on 3D surface model by deep learning,” IEEE Access , vol. 9, pp. 157 628–157 638, 2021

2021
[26]

Evaluation of a novel three-dimensional wound measurement device for assessment of diabetic foot ulcers,

G. Pena, B. Kuang, Z. Szpak, P. Cowled, J. Dawson, and R. Fitridge, “Evaluation of a novel three-dimensional wound measurement device for assessment of diabetic foot ulcers,” Adv. Wound Care, vol. 9, no. 11, pp. 623–631, 2020

2020
[27]

Integrating 3D model representation for an accurate non-invasive assessment of pressure injuries with deep learning,

S. Zahia, B. Garcia-Zapirain, and A. Elmaghraby, “Integrating 3D model representation for an accurate non-invasive assessment of pressure injuries with deep learning,” Sensors, vol. 20, no. 10, p. 2933, 2020

2020
[28]

RGB-D camera-based automatic wound- measurement system,

P. Zhang, Y . Zhang, and Q. Li, “RGB-D camera-based automatic wound- measurement system,” IEEE Trans. Instrum. Meas. , vol. 72, pp. 1–11, 2023

2023
[29]

Multimodal sensor system for pressure ulcer wound assessment and care,

M.-C. Chang, T. Yu, J. Luo, K. Duan, P. Tu, Y . Zhao, N. Nagraj, V . Rajiv, M. Priebe, E. A. Wood et al. , “Multimodal sensor system for pressure ulcer wound assessment and care,” IEEE Trans. Ind. Informatics , vol. 14, no. 3, pp. 1186–1196, 2017

2017
[30]

Develop- ment of a robust photogrammetric metrology system for monitoring the healing of bedsores,

A. Malian, A. Azizi, F. A. van den Heuvel, and M. Zolfaghari, “Develop- ment of a robust photogrammetric metrology system for monitoring the healing of bedsores,” Photogramm. Rec. , vol. 20, no. 111, pp. 241–273, 2005

2005
[31]

Enhanced assessment of the wound-healing process by accurate multiview tissue classification,

H. Wannous, Y . Lucas, and S. Treuillet, “Enhanced assessment of the wound-healing process by accurate multiview tissue classification,” IEEE Trans. Med. Imaging , vol. 30, no. 2, pp. 315–326, 2010

2010
[32]

Autonomous robot-driven chronic wound 3D reconstruction and analysis system,

D. Filko and E. K. Nyarko, “Autonomous robot-driven chronic wound 3D reconstruction and analysis system,” Robotics, vol. 14, no. 3, p. 30, 2025

2025
[33]

System design for 3D wound imaging using low-cost mobile devices,

E. Sirazitdinova and T. M. Deserno, “System design for 3D wound imaging using low-cost mobile devices,” in Proc. SPIE Med. Imaging: Imaging Informatics for Healthcare, Research, and Appl. , vol. 10138, 2017, pp. 258–264

2017
[34]

Medetec wound database,

S. Thomas, “Medetec wound database,” 2020

2020
[35]

Fully automatic wound segmentation with deep convolutional neural networks,

C. Wang, D. Anisuzzaman, V . Williamson, M. K. Dhar, B. Rostami, J. Niezgoda, S. Gopalakrishnan, and Z. Yu, “Fully automatic wound segmentation with deep convolutional neural networks,” Sci. Rep., vol. 10, no. 1, p. 21897, 2020

2020
[36]

Fuseg: The foot ulcer segmentation challenge,

C. Wang, A. Mahbod, I. Ellinger, A. Galdran, S. Gopalakrishnan, J. Niezgoda, and Z. Yu, “Fuseg: The foot ulcer segmentation challenge,” arXiv preprint arXiv:2201.00414 , 2022

2022 arXiv
[37]

Translating clinical delineation of diabetic foot ulcers into machine interpretable segmentation,

C. Kendrick, B. Cassidy, J. M. Pappachan, C. O’Shea, C. J. Fernandez, E. Chacko, K. Jacob, N. D. Reeves, and M. H. Yap, “Translating clinical delineation of diabetic foot ulcers into machine interpretable segmentation,” arXiv preprint arXiv:2204.11618 , 2022

2022 arXiv
[38]

ComplexWoundDB: A database for automatic complex wound tissue categorization,

T. A. Pereira, R. C. Popim, L. A. Passos, D. R. Pereira, C. R. Pereira, and J. P. Papa, “ComplexWoundDB: A database for automatic complex wound tissue categorization,” in Proc. Int. Conf. Syst., Signals Image Process. (IWSSIP), 2022, pp. 1–4. 12

2022
[39]

Chronic wounds multimodal image database,

M. Kr˛ ecichwost, J. Czajkowska, A. Wijata, J. Juszczyk, B. Pyci ´nski, M. Biesok, M. Rudzki, J. Majewski, J. Kostecki, and E. Pietka, “Chronic wounds multimodal image database,” Comput. Med. Imaging Graph. , vol. 88, p. 101844, 2021

2021
[40]

DermSynth3D: Synthesis of in-the-wild annotated dermatology images,

A. Sinha, J. Kawahara, A. Pakzad, K. Abhishek, M. Ruthven, E. Ghorbel, A. Kacem, D. Aouada, and G. Hamarneh, “DermSynth3D: Synthesis of in-the-wild annotated dermatology images,” Med. Image Anal. , vol. 95, p. 103145, 2024

2024
[41]

Syn3DWound: A synthetic dataset for 3D wound bed analysis,

L. Lebrat, R. Santa Cruz, R. Chierchia, Y . Arzhaeva, M. A. Armin, J. Goldsmith, J. Oorloff, P. Reddy, C. Nguyen, L. Petersson et al. , “Syn3DWound: A synthetic dataset for 3D wound bed analysis,” in Proc. IEEE Int. Symp. Biomed. Imaging (ISBI) , 2024, pp. 1–5

2024
[42]

SALVE: A 3D Reconstruction Benchmark of Wounds from Consumer-grade Videos,

R. Chierchia, L. Lebrat, D. Ahmedt-Aristizabal, O. Salvado, C. Fookes, and R. S. Cruz, “SALVE: A 3D Reconstruction Benchmark of Wounds from Consumer-grade Videos,” in Proc. IEEE Winter Conf. Appl. Comput. Vis. (WACV), 2025

2025
[43]

Measuring surface area of skin lesions with 2D and 3D algorithms,

H. Mirzaalian Dastjerdi, D. Töpfer, S. J. Rupitsch, and A. Maier, “Measuring surface area of skin lesions with 2D and 3D algorithms,” Int. J. Biomed. Imaging , vol. 2019, no. 1, p. 4035148, 2019

2019
[44]

Wound area measurement with 3D transformation and smartphone images,

C. Liu, X. Fan, Z. Guo, Z. Mo, E. I.-C. Chang, and Y . Xu, “Wound area measurement with 3D transformation and smartphone images,” BMC Bioinform., vol. 20, pp. 1–21, 2019

2019
[45]

SfM-3DULC: Reliability of a new 3D wound measurement procedure and its accuracy in projected area,

D. Sánchez-Jiménez, F. F. Buchón-Moragues, B. Escutia-Muñoz, and R. Botella-Estrada, “SfM-3DULC: Reliability of a new 3D wound measurement procedure and its accuracy in projected area,” Int. Wound J., vol. 19, no. 1, pp. 44–51, 2022

2022
[46]

Reliability of a three-dimensional wound camera and correlation with routine ruler measurement in diabetes-related foot ulceration,

J. W. J. Lasschuit, J. Featherston, and K. T. T. Tonks, “Reliability of a three-dimensional wound camera and correlation with routine ruler measurement in diabetes-related foot ulceration,” J. Diabetes Sci. Technol., vol. 15, no. 6, pp. 1361–1367, 2021

2021
[47]

Wound perimeter, area, and volume measurement based on laser 3D and color acquisition,

U. Pavlovˇciˇc, J. Diaci, J. Možina, and M. Jezeršek, “Wound perimeter, area, and volume measurement based on laser 3D and color acquisition,” BioMed. Eng. Online , vol. 14, pp. 1–15, 2015

2015
[48]

SegFormer: Simple and efficient design for semantic segmentation with transformers,

E. Xie, W. Wang, Z. Yu et al. , “SegFormer: Simple and efficient design for semantic segmentation with transformers,” Adv. Neural Inf. Process. Syst., vol. 34, pp. 12 077–12 090, 2021

2021
[49]

MA VIS: a non-invasive instrument to measure area and volume of wounds,

P. Plassmann and T. Jones, “MA VIS: a non-invasive instrument to measure area and volume of wounds,” Med. Eng. Phys., vol. 20, no. 5, pp. 332–338, 1998

1998
[50]

AliceVision Meshroom: An open-source 3D reconstruction pipeline,

C. Griwodz, S. Gasparini, L. Calvet, P. Gurdjos, F. Castan, B. Maujean, G. De Lillo, and Y . Lanthony, “AliceVision Meshroom: An open-source 3D reconstruction pipeline,” in ACM Multimed. Syst. Conf , 2021

2021
[51]

A parallel algorithm for polygon rasterization,

J. Pineda, “A parallel algorithm for polygon rasterization,” in Proc. Annu. Conf. Comput. Graph. Interact. Tech. , 1988, pp. 17–20

1988
[52]

Stability of weighted majority voting under estimated weights,

S. Bai, D. Wang, T. Muller, P. Cheng, and J. Chen, “Stability of weighted majority voting under estimated weights,” in Proc. Int. Conf. Auton. Agents Multiagent Syst. (AAMAS) , 2024, pp. 96–104

2024
[53]

A graph-based mathematical morphology reader,

L. Najman and J. Cousty, “A graph-based mathematical morphology reader,” Pattern Recognit. Lett. , vol. 47, pp. 3–17, 2014

2014
[54]

Hartley and A

R. Hartley and A. Zisserman, Multiple view geometry in computer vision . Cambridge university press, 2003

2003
[55]

Reliability of electronic versus manual wound measurement techniques,

S. Haghpanah, K. Bogie, X. Wang, P. G. Banks, and C. H. Ho, “Reliability of electronic versus manual wound measurement techniques,” Arch. Phys. Med. Rehabil. , vol. 87, no. 10, pp. 1396–1402, 2006

2006
[56]

Automatic robot-driven 3d reconstruction system for chronic wounds,

D. Filko, D. Marijanovi ´c, and E. K. Nyarko, “Automatic robot-driven 3d reconstruction system for chronic wounds,” Sensors, vol. 21, no. 24, p. 8308, 2021

2021
[57]

A study of the efficiency and convenience of an advanced portable wound measurement system (visitraktm),

J. Sugama, Y . Matsui, H. Sanada, C. Konya, M. Okuwa, and A. Kitagawa, “A study of the efficiency and convenience of an advanced portable wound measurement system (visitraktm),” Journal of clinical nursing , vol. 16, no. 7, pp. 1265–1269, 2007

2007
[58]

Wound measurement: can it help us to monitor progression to healing?

M. Flanagan, “Wound measurement: can it help us to monitor progression to healing?” Journal of wound care , vol. 12, no. 5, pp. 189–194, 2003

2003
[59]

Generative ai in medical imaging: applications, challenges, and ethics,

M. Koohi-Moghadam and K. T. Bae, “Generative ai in medical imaging: applications, challenges, and ethics,” J. Med. Syst. , vol. 47, no. 1, p. 94, 2023

2023
[60]

Deblur-nerf: Neural radiance fields from blurry images,

L. Ma, X. Li, J. Liao, Q. Zhang, X. Wang, J. Wang, and P. V . Sander, “Deblur-nerf: Neural radiance fields from blurry images,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), June 2022, pp. 12 861–12 870

2022

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.

[1] [23]

Wound measurement by RGB-D camera,

D. Filko, R. Cupec, and E. K. Nyarko, “Wound measurement by RGB-D camera,” Mach. Vis. Appl. , vol. 29, pp. 633–654, 2018

work page 2018

[2] [24]

Three-dimensional wound reconstruction using point descriptors: A comparative study,

J. R. Souto, F. M. Barbosa, and B. M. Carvalho, “Three-dimensional wound reconstruction using point descriptors: A comparative study,” in Proc. IEEE Int. Symp. Comput.-Based Med. Syst. (CBMS) , 2023, pp. 41–46

work page 2023

[3] [1]

It was curated and used to train and validate our 2D segmentation model, a core component of the 3D segmentation pipeline

Retrospective data: This dataset includes 2D wound images captured in simulated clinical settings and shared under a data use agreement. It was curated and used to train and validate our 2D segmentation model, a core component of the 3D segmentation pipeline. The images were not used for 3D reconstruction or measurement tasks

work page

[4] [2]

Videos were recorded using iPhones and Logitech cameras, while a Revopoint 3D scanner provided geometric ground truth

Prospective data: This dataset is collected in collabora- tion with clinical researchers, following a structured acquisition protocol to ensure consistency and suitability for 3D recon- struction, to enable real-world validation. Videos were recorded using iPhones and Logitech cameras, while a Revopoint 3D scanner provided geometric ground truth. The whol...

work page

[5] [3]

By directly operating on the mesh structure and leveraging surface triangulation, this method provides precise and robust estimates

Wound bed and periwound surface area: The wound bed and periwound areas are computed by summing the areas of triangular mesh faces assigned to each respective segmentation label in the reconstructed surface. By directly operating on the mesh structure and leveraging surface triangulation, this method provides precise and robust estimates

work page

[6] [4]

These frames are critical for accurately reconstructing the wound’s shape and surrounding tissue

Pre-processing: To create a high-quality 3D wound model, our framework selects a set of sharp frames that capture the wound from multiple angles. These frames are critical for accurately reconstructing the wound’s shape and surrounding tissue. Typically, � � �� frames are used, which provides a good balance between detail coverage and visual quality. Whil...

work page

[7] [5]

3D reconstruction methods: Currently, our framework adopts the Meshroom pipeline [32] for 3D reconstruction. Specifically, we applied two custom settings to optimize for our dataset: i) Single-Camera Model : Tailored for monocular video input; ii) High-Density SIFT Features: Up to 50,000 SIFT features per image to improve detail, with a minor increase in ...

work page

[8] [6]

2D segmentation: Our method builds upon the SegFormer architecture [30], specifically the MiT-b5 variant, which we fine- tuned to predict 2D segmentations of the wound bed, periwound region, and four tissue classes—granulation, necrotic, slough, and epithelial. Specifically, we fine-tuned the model for three distinct tasks: binary segmentation of the woun...

work page