Classification of racehorse limb radiographs using deep convolutional neural networks.
Abstract: To assess the capability of deep convolutional neural networks to classify anatomical location and projection from a series of 48 standard views of racehorse limbs. Unassigned: Radiographs ( = 9504) of horse limbs from image sets made for veterinary inspections by 10 independent veterinary clinics were used to train, validate and test (116, 40 and 42 radiographs, respectively) six deep learning architectures available as part of the open source machine learning framework PyTorch. The deep learning architectures with the best top-1 accuracy had the batch size further investigated. Unassigned: Top-1 accuracy of six deep learning architectures ranged from 0.737 to 0.841. Top-1 accuracy of the best deep learning architecture (ResNet-34) ranged from 0.809 to 0.878, depending on batch size. ResNet-34 (batch size = 8) achieved the highest top-1 accuracy (0.878) and the majority (91.8%) of misclassification was due to laterality error. Class activation maps indicated that joint morphology, not side markers or other non-anatomical image regions, drove the model decision. Unassigned: Deep convolutional neural networks can classify equine pre-import radiographs into the 48 standard views including moderate discrimination of laterality, independent of side marker presence.
© 2023 The Authors. Veterinary Record Open published by John Wiley & Sons Ltd on behalf of British Veterinary Association.
Publication Date: 2023-01-29 PubMed ID: 36726400PubMed Central: PMC9884469DOI: 10.1002/vro2.55Google Scholar: Lookup
The Equine Research Bank provides access to a large database of publicly available scientific literature. Inclusion in the Research Bank does not imply endorsement of study methods or findings by Mad Barn.
- Journal Article
Summary
This research summary has been generated with artificial intelligence and may contain errors and omissions. Refer to the original study to confirm details provided. Submit correction.
This research article investigates how deep convolutional neural networks perform in classifying standard views of racehorse limb radiographs. The networks were trained with nearly 10,000 radiographs from ten veterinary clinics and the most effective architecture was found to be ResNet-34.
Research Methodology
- The researchers used a collection of 9,504 radiographs of horse limbs. These images were collected from different sets made for veterinary inspections by 10 independent veterinary clinics.
- The collected radiographs were used to train, test, and validate six different deep learning architectures. These architectures were part of PyTorch, an open-source machine learning framework.
- The research examined the impact of the batch size on the top-1 accuracy of the deep learning architectures. The best batch size was selected for the architecture with the highest top-1 accuracy.
Research Findings
- The top-1 accuracy rate of the six deep learning architectures ranged from 0.737 to 0.841.
- ResNet-34 was found to be the best deep learning architecture, with its top-1 accuracy rate varying between 0.809 and 0.878, depending on the batch size.
- ResNet-34 achieved the highest top-1 accuracy (0.878) when the batch size was 8.
- A significant portion (91.8%) of the misclassification was found to be due to errors in distinguishing laterality (whether the radiograph was of the left or right limb).
- Through examination of class activation maps, it was discovered that the model decisions were primarily driven by joint morphology, rather than side markers or other non-anatomical image regions.
Conclusion
- Overall, the study concludes that deep convolutional neural networks can classify equine pre-import radiographs into the 48 standard views, with a moderate ability to discriminate laterality, regardless of the presence of side markers.
Cite This Article
APA
Costa da Silva RG, Mishra AP, Riggs CM, Doube M.
(2023).
Classification of racehorse limb radiographs using deep convolutional neural networks.
Vet Rec Open, 10(1), e55.
https://doi.org/10.1002/vro2.55 Publication
Researcher Affiliations
- Department of Infectious Diseases and Public Health City University of Hong Kong Hong Kong SAR China.
- Department of Infectious Diseases and Public Health City University of Hong Kong Hong Kong SAR China.
- Department of Veterinary Clinical Services Hong Kong Jockey Club Hong Kong SAR China.
- Department of Infectious Diseases and Public Health City University of Hong Kong Hong Kong SAR China.
Conflict of Interest Statement
The authors declare they have no conflicts of interest.
References
This article includes 32 references
- Çallı E, Sogancioglu E, van Ginneken B, van Leeuwen KG, Murphy K. Deep learning for chest X-ray analysis: A survey.. Med Image Anal 2021 Aug;72:102125.
- Mohammad-Rahimi H, Nadimi M, Rohban MH, Shamsoddin E, Lee VY, Motamedian SR. Machine learning and orthodontics, current trends and the future opportunities: A scoping review.. Am J Orthod Dentofacial Orthop 2021 Aug;160(2):170-192.e4.
- Litjens G, Kooi T, Bejnordi BE, Setio AAA, Ciompi F, Ghafoorian M, van der Laak JAWM, van Ginneken B, Sánchez CI. A survey on deep learning in medical image analysis.. Med Image Anal 2017 Dec;42:60-88.
- Yu KH, Beam AL, Kohane IS. Artificial intelligence in healthcare.. Nat Biomed Eng 2018 Oct;2(10):719-731.
- Hirschmann A, Cyriac J, Stieltjes B, Kober T, Richiardi J, Omoumi P. Artificial Intelligence in Musculoskeletal Imaging: Review of Current Literature, Challenges, and Trends.. Semin Musculoskelet Radiol 2019 Jun;23(3):304-311.
- D'Angelo T, Caudo D, Blandino A, Albrecht MH, Vogl TJ, Gruenewald LD, Gaeta M, Yel I, Koch V, Martin SS, Lenga L, Muscogiuri G, Sironi S, Mazziotti S, Booz C. Artificial intelligence, machine learning and deep learning in musculoskeletal imaging: Current applications.. J Clin Ultrasound 2022 Nov;50(9):1414-1431.
- Cihan P, Gökçe E, Kalipsiz O. A review of machine learning applications in veterinary field. Kafkas Univ Vet Fak Derg 2017;23(4):673–80.
- Gomes DA, Alves‐Pimenta MS, Ginja M, Filipe V. Predicting canine hip dysplasia in x‐ray images using deep learning. In: Pereira AI, Fernandes FP, Coelho JP, Teixeira JP, Pacheco MF, Alves P, et al., editors. Optimization, learning algorithms and applications (Communications in Computer and Information Science, volume 1488; Proceedings of the first OL2A conference, Portugal, 2021). Cham, Switzerland: Springer International Publishing; 2021. p. 393–400.
- Basran PS, Appleby RB. The unmet potential of artificial intelligence in veterinary medicine.. Am J Vet Res 2022 Mar 30;83(5):385-392.
- Olstad K, Gangsei LE, Kongsro J. A method for labelling lesions for machine learning and some new observations on osteochondrosis in computed tomographic scans of four pig joints.. BMC Vet Res 2022 Aug 31;18(1):328.
- Rizzo S, Botta F, Raimondi S, Origgi D, Fanciullo C, Morganti AG, Bellomi M. Radiomics: the facts and the challenges of image analysis.. Eur Radiol Exp 2018 Nov 14;2(1):36.
- Basran PS, Gao J, Palmer S, Reesink HL. A radiomics platform for computing imaging features from µCT images of Thoroughbred racehorse proximal sesamoid bones: Benchmark performance and evaluation.. Equine Vet J 2021 Mar;53(2):277-286.
- Awaysheh A, Wilcke J, Elvinger F, Rees L, Fan W, Zimmerman KL. Review of Medical Decision Support and Machine-Learning Methods.. Vet Pathol 2019 Jul;56(4):512-525.
- Rajpurkar P, Irvin J, Bagul A, Ding D, Duan T, Mehta H. MURA: large dataset for abnormality detection in musculoskeletal radiographs. arXiv 2017; 1712.06957.
- . The Hong Kong Jockey Club veterinary pre‐import examination protocol. Hong Kong: Department of Veterinary Regulation, Welfare and Biosecurity Policy; 2020.
- Solounias N, Danowitz M, Stachtiaris E, Khurana A, Araim M, Sayegh M, Natale J. The evolution and anatomy of the horse manus with an emphasis on digit reduction.. R Soc Open Sci 2018 Jan;5(1):171782.
- Rosanowski SM, Chang YM, Stirk AJ, Verheyen KLP. Epidemiology of race-day distal limb fracture in flat racing Thoroughbreds in Great Britain (2000-2013).. Equine Vet J 2019 Jan;51(1):83-89.
- Canziani A, Paszke A, Culurciello E. An analysis of deep neural network models for practical applications. arXiv 2017; 1605.07678.
- Ananda A, Ngan KH, Karabağ C, Ter-Sarkisov A, Alonso E, Reyes-Aldasoro CC. Classification and Visualisation of Normal and Abnormal Radiographs; A Comparison between Eleven Convolutional Neural Network Architectures.. Sensors (Basel) 2021 Aug 9;21(16).
- Paszke A, Gross S, Massa F, Lerer A, Bradbury J, Chanan G. PyTorch: an imperative style, high‐performance deep learning library. In: NIPS'19: Proceedings of the 33rd International Conference on Neural Information Processing Systems Red Hook, NY, USA: Curran Associates Inc.; 2019. p. 8026–37. Accessed December 12, 2022.
- Zhou B, Khosla A, Lapedriza A, Oliva A, Torralba A. Learning deep features for discriminative localization. arXiv 2015.
- Olczak J, Fahlberg N, Maki A, Razavian AS, Jilert A, Stark A, Sköldenberg O, Gordon M. Artificial intelligence for analyzing orthopedic trauma radiographs.. Acta Orthop 2017 Dec;88(6):581-586.
- Filice RW, Frantz SK. Effectiveness of Deep Learning Algorithms to Determine Laterality in Radiographs.. J Digit Imaging 2019 Aug;32(4):656-664.
- Roy TA, Ruff CB, Plato CC. Hand dominance and bilateral asymmetry in the structure of the second metacarpal.. Am J Phys Anthropol 1994 Jun;94(2):203-11.
- Leśniak KG, Williams JM. Relationship Between Magnitude and Direction of Asymmetries in Facial and Limb Traits in Horses and Ponies.. J Equine Vet Sci 2020 Oct;93:103195.
- Leśniak K. The incidence of, and relationship between, distal limb and facial asymmetry, and performance in the event horse. Comp Exerc Physiol 2020;16(1):47–53.
- Watson KM, Stitson DJ, Davies HM. Third metacarpal bone length and skeletal asymmetry in the Thoroughbred racehorse.. Equine Vet J 2003 Nov;35(7):712-4.
- Pearce GP, May-Davis S, Greaves D. Femoral asymmetry in the Thoroughbred racehorse.. Aust Vet J 2005 Jun;83(6):367-70.
- Masters D, Luschi C. Revisiting small batch training for deep neural networks. arXiv 2018.
- Saun TJ. Automated Classification of Radiographic Positioning of Hand X-Rays Using a Deep Neural Network.. Plast Surg (Oakv) 2021 May;29(2):75-80.
- Drenkow N, Sani N, Shpitser I, Unberath M. A systematic review of robustness in deep learning for computer vision: mind the gap?. arXiv 2022.
- Fang X, Harris L, Zhou W, Huo D. Generalized Radiographic View Identification with Deep Learning.. J Digit Imaging 2021 Feb;34(1):66-74.
Citations
This article has been cited 0 times.Use Nutrition Calculator
Check if your horse's diet meets their nutrition requirements with our easy-to-use tool Check your horse's diet with our easy-to-use tool
Talk to a Nutritionist
Discuss your horse's feeding plan with our experts over a free phone consultation Discuss your horse's diet over a phone consultation
Submit Diet Evaluation
Get a customized feeding plan for your horse formulated by our equine nutritionists Get a custom feeding plan formulated by our nutritionists
