mlsanityData to 7 Oct 2026

An Enhanced Harmonic Densely Connected Hybrid Transformer Network Architecture for Chronic Wound Segmentation Utilising Multi-Colour Space Tensor Merging

arXiv 2410.03359 ↗First posted 4 Oct 2024eess.IV cs.AI cs.CVcited by 2

Not on the map. Why?

Abstract (arXiv)

Chronic wounds and associated complications present ever growing burdens for clinics and hospitals world wide. Venous, arterial, diabetic, and pressure wounds are becoming increasingly common globally. These conditions can result in highly debilitating repercussions for those affected, with limb amputations and increased mortality risk resulting from infection becoming more common. New methods to assist clinicians in chronic wound care are therefore vital to maintain high quality care standards. This paper presents an improved HarDNet segmentation architecture which integrates a contrast-eliminating component in the initial layers of the network to enhance feature learning. We also utilise a multi-colour space tensor merging process and adjust the harmonic shape of the convolution blocks to facilitate these additional features. We train our proposed model using wound images from light-skinned patients and test the model on two test sets (one set with ground truth, and one without) comprising only darker-skinned cases. Subjective ratings are obtained from clinical wound experts with intraclass correlation coefficient used to determine inter-rater reliability. For the dark-skin tone test set with ground truth, we demonstrate improvements in terms of Dice similarity coefficient (+0.1221) and intersection over union (+0.1274). Qualitative analysis showed high expert ratings, with improvements of >3% demonstrated when comparing the baseline model with the proposed model. This paper presents the first study to focus on darker-skin tones for chronic wound segmentation using models trained only on wound images exhibiting lighter skin. Diabetes is highly prevalent in countries where patients have darker skin tones, highlighting the need for a greater focus on such cases. Additionally, we conduct the largest qualitative study to date for chronic wound segmentation.

Key facts. Cited by 2 papers in the map (data through 7 Oct 2026). Builds on 31 of its 110 references that are in the map. First posted on arXiv 4 Oct 2024.

Builds on

31 of 110 references.

References · 31

  1. 2015U-Net: Convolutional Networks for Biomedical Image Segmentationcited by 19,604
  2. 2015Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shiftcited by 6,979
  3. 2015Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classificationcited by 3,590
  4. 2021MLP-Mixer: An all-MLP Architecture for Visioncited by 1,441
  5. 2016Instance Normalization: The Missing Ingredient for Fast Stylizationcited by 1,187
  6. 2014Spatial Pyramid Pooling in Deep Convolutional Networks for Visual Recognitioncited by 719
  7. 2018Albumentations: fast and flexible image augmentationscited by 558
  8. 2014Deeply-Supervised Netscited by 364
  9. 2018Two at Once: Enhancing Learning and Generalization Capacities via IBN-Netcited by 361
  10. 2019ResUNet++: An Advanced Architecture for Medical Image Segmentationcited by 211
  11. 2022UNeXt: MLP-based Rapid Medical Image Segmentation Networkcited by 185
  12. 2021Evaluating Deep Neural Networks Trained on Clinical Images in Dermatology with the Fitzpatrick 17k Datasetcited by 184
19 more
  1. 2019The VIA Annotation Software for Images, Audio and Videocited by 166
  2. 2021HarDNet-MSEG: A Simple Encoder-Decoder Polyp Segmentation Neural Network that Achieves over 0.9 Mean Dice and 86 FPScited by 71
  3. 2022DCSAU-Net: A Deeper and More Compact Split-Attention U-Net for Medical Image Segmentationcited by 37
  4. 2022How to Train Vision Transformer on Small-scale Datasets?cited by 34
  5. 2020Fully Automatic Wound Segmentation with Deep Convolutional Neural Networkscited by 32
  6. 2019ColorNet: Investigating the importance of color spaces for image classificationcited by 31
  7. 2023Understanding Why ViT Trains Badly on Small Datasets: An Intuitive Perspectivecited by 28
  8. 2022FUSeg: The Foot Ulcer Segmentation Challengecited by 21
  9. 2020Deep Learning in Diabetic Foot Ulcers Detection: A Comprehensive Evaluationcited by 19
  10. 2022Translating Clinical Delineation of Diabetic Foot Ulcers into Machine Interpretable Segmentationcited by 19
  11. 2017Fully Convolutional Networks for Diabetic Foot Ulcer Segmentationcited by 17
  12. 2019Switchable Normalization for Learning-to-Normalize Deep Representationcited by 16
  13. 2021Analysis Towards Classification of Infection and Ischaemia of Diabetic Foot Ulcerscited by 16
  14. 2021Detect-and-Segment: a Deep Learning Approach to Automate Wound Image Segmentationcited by 12
  15. 2021Automatic Foot Ulcer Segmentation Using an Ensemble of Convolutional Neural Networkscited by 9
  16. 2021A Cloud-based Deep Learning Framework for Remote Detection of Diabetic Foot Ulcerscited by 7
  17. 2022HarDNet-DFUS: An Enhanced Harmonically-Connected Network for Diabetic Foot Ulcer Image Segmentation and Colonoscopy Polyp Segmentationcited by 6
  18. 2022ComplexWoundDB: A Database for Automatic Complex Wound Tissue Categorizationcited by 3
  19. 2023Dermoscopic Dark Corner Artifacts Removal: Friend or Foe?cited by 2

Built on by

Cited by year (2026 to 7 Oct)

Citing papers per year: 2025: 1; 2026: 1.

2 papers.

  1. 2025Gaussian Random Fields as an Abstract Representation of Patient Metadata for Multimodal Medical Image Segmentationcited by 0
  2. 2026Automated multi-class wound assessment using dedicated instance segmentation models for boundary detection and classificationcited by 0