Performance of convolutional neural networks in the histopathological grading of canine cutaneous mast cell tumors: a pilot study
DOI:
https://doi.org/10.36560/19620262291Palavras-chave:
Dermatopathology; Artificial intelligence; Veterinary pathologyResumo
Canine cutaneous mast cell tumor is one of the most prevalent skin neoplasms in dogs, and histological grading is essential for prognosis and therapeutic decision‑making. However, histological classification according to the Patnaik and Kiupel systems presents limitations related to subjectivity and low reproducibility among pathologists. This study evaluated the performance of convolutional neural networks (CNNs) in the histopathological classification of canine cutaneous mast cell tumors. Eighteen histological slides labeled by specialists were used and processed into image patches. The AlexNet, InceptionV3, ResNet50, and YOLO11 architectures were trained using cross‑validation and evaluated based on accuracy, sensitivity, precision, F1‑score, and area under the curve (AUC). Grad‑CAM techniques and feature maps were also applied for interpretative analysis of the models’ decisions. The YOLO11 and ResNet50 models showed the best performance, achieving AUC values of up to 93.5%, but did not reach the arbitrarily defined minimum clinical threshold of 95%. Visual analysis demonstrated that the models recognized relevant morphological patterns, such as granules and pleomorphism, but failed to differentiate between grades II and III. It is concluded that, although promising as auxiliary tools, CNNs still require methodological improvements, such as refined curation, semantic segmentation, and cellular structure identification, for clinical application.
Referências
BAND, Shahab S. et al. Application of explainable artificial intelligence in medical health: A systematic review of interpretability methods. Informatics in Medicine Unlocked, v.40, p.101286, 2023.
BERTRAM, C. A. et al. MITOS_WSI_CCMCT: a dataset for mitotic figure assessment in canine cutaneous mast cell tumors. Veterinary Pathology, v.56, n.3, p.393–402, 2019. DOI: https://doi.org/10.1177/0300985818815317.
BERTRAM, Christof A. et al. Are pathologist-defined labels reproducible? Comparison of the TUPAC16 mitotic figure dataset with an alternative set of labels. In: Interpretable and Annotation-Efficient Learning for Medical Image Computing: MICCAI 2020. Cham: Springer, 2020. p.204–213.
BLACKWOOD, L. et al. European consensus document for mast cell tumours in dogs and cats. Veterinary and Comparative Oncology, v.18, n.3, p.439–452, 2020. DOI: https://doi.org/10.1111/vco.12665.
BOMMASANI, R. et al. On the opportunities and risks of foundation models. arXiv preprint, arXiv:2108.07258, 2021.
CHEN, Wuyang; GONG, Xinyu; WANG, Zhangyang. Neural architecture search on imagenet in four gpu hours: A theoretically inspired perspective. arXiv preprint arXiv:2102.11535, 2021.
CHRISTOF, A. et al. Challenges in applying deep learning to histopathological images in veterinary medicine. Veterinary Pathology Insights, v. 13, p. 1–10, 2020.
DE NARDI, A. B. et al. Diagnosis, Prognosis and Treatment of Canine Cutaneous and Subcutaneous Mast Cell Tumors. Cells, v. 11, n. 4, p. 618, jan. 2022.
DE NARDI, A. B. et al. Tumores cutâneos em cães: uma revisão de 10 anos. Veterinária e Zootecnia, v. 29, n. 1, p. 10–20, 2022.
ELFER, K. et al. Pilot study to evaluate tools to collect pathologist annotations for validating machine learning algorithms. Journal of Medical Imaging (Bellingham), v.9, n.4, p.047501, jul. 2022. DOI: 10.1117/1.JMI.9.4.047501.
FURLANI, J. M. et al. Mastocitoma canino: estudo retrospectivo. Ciência Animal Brasileira, v. 9, n. 1, p. 242–250, 2008.
FURLANI, R. P. Z. et al. Mastocitoma cutâneo canino: aspectos clínicos, diagnósticos e terapêuticos. Ciência Rural, v.38, n.4, p.1132–1138, 2008. DOI: https://doi.org/10.1590/S0103-84782008000400038.
GROSS, T. L. et al. Skin diseases of the dog and cat: clinical and histopathologic diagnosis. 2. ed. Oxford: Blackwell Publishing, 2005.
JIAO, Y. et al. A semi-supervised deep learning framework for histopathology image analysis in low-data regimes. Medical Image Analysis, v. 85, 102747, 2024. DOI: https://doi.org/10.1016/j.media.2024.102747.
KHANAM, R.; HUSSAIN, M. YOLOv11: An Overview of the Key Architectural Enhancements. arXiv preprint, arXiv:2410.17725, 2024.
KIUPEL, M. et al. Proposal of a 2-tier histologic grading system for canine cutaneous mast cell tumors to replace the Patnaik system. Veterinary Pathology, v. 48, n. 1, p. 147–155, 2011. DOI: https://doi.org/10.1177/0300985810386469.
LU, M. Y. et al. Data-efficient and weakly supervised computational pathology on whole-slide images. Nature Biomedical Engineering, v. 5, p. 555–570, 2021. DOI: https://doi.org/10.1038/s41551-020-00682-w.
MAHLICH, L.; WILHELM, F.; HOPPE, T. Efficient e-Fold Cross-Validation for Deep Neural Networks. arXiv preprint, arXiv:2401.13416, 2024.
MCGENITY, Clare et al. Artificial intelligence in digital pathology: a systematic review and meta-analysis of diagnostic test accuracy. npj Digital Medicine, v. 7, n. 1, p. 114, 2024.
MEUTEN, D. J. Tumors in Domestic Animals. 4. ed. Iowa: Iowa State Press, 2017.
MISDORP, W. et al. Histological classification of mammary tumors of the dog and cat. 2. ed. Washington, DC: Armed Forces Institute of Pathology, 1999.
MONTANARI, M. F. Técnicas de microscopia óptica aplicadas à histopatologia veterinária. In: MEUTEN, D. J. (Org.). Patologia veterinária especial. 5. ed. São Paulo: Roca, 2016. p. 31–45.
NORVIG, P. Inteligência Artificial. 4. ed. Rio de Janeiro: Grupo Gen, 2022.
PASZKE, A. et al. PyTorch: An Imperative Style, High-Performance Deep Learning Library. arXiv preprint, arXiv:1912.01703, 2019.
PATNAIK, A. K. et al. Canine cutaneous mast cell tumor: morphologic grading and survival time in 83 dogs. Veterinary Pathology, v. 21, n. 5, p. 469–474, 1984.
PEDREGOSA, F. et al. Scikit-learn: Machine Learning in Python. Journal of Machine Learning Research, v. 12, p. 2825–2830, 2011.
PETRÍKOVÁ, I. et al. A deep learning approach for histopathological image classification in veterinary oncology. Computers in Biology and Medicine, v. 158, 106957, 2023. DOI: https://doi.org/10.1016/j.compbiomed.2023.106957.
PRADO, Aline Ambrogi Franco et al. Mastocitoma em cães: Aspectos clínicos, histopatológicos e tratamento. Enciclopédia Biosfera, Centro Científico Conhecer, v. 8, n. 14, p. 2151, 2012.
RAMOS-VARA, J. A.; MILLER, M. A. When tissue antigens and antibodies get along: revisiting the technical aspects of immunohistochemistry—the red, brown, and blue technique. Veterinary pathology, v. 51, n. 1, p. 42-87, 2014.
RIDNIK, T. et al. Imagenet pretraining: should we train from scratch? In: Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), 2021. p. 9224–9233.
RÍOS, R. M. Mastócitos: funções biológicas e implicações patológicas. Revista de Imunologia Veterinária, v. 13, n. 2, p. 45–52, 2008.
SALVI, M. et al. Histopathological classification of canine cutaneous round cell tumors using deep learning: a multi‑center study. Frontiers in Veterinary Science, v. 8, p. 640944, 26 mar. 2021. DOI: 10.3389/fvets.2021.640944
SELVARAJU, R. R. et al. Grad‑CAM: Visual Explanations from Deep Networks via Gradient‑Based Localization. Proceedings of the IEEE International Conference on Computer Vision (ICCV), Venice, Italy, p. 618–626, 2017. DOI: 10.1109/ICCV.2017.74
SIMÕES, D. M. et al. Variabilidade interobservador na classificação histológica de mastocitomas caninos. Arquivo Brasileiro de Medicina Veterinária e Zootecnia, v. 46, n. 3, p. 327–334, 1994.
SPRINGENBERG, Jost Tobias et al. Striving for simplicity: The all convolutional net. arXiv preprint arXiv:1412.6806, 2014. DOI: https://doi.org/10.48550/arXiv.1412.6806
WANG, Yaqing et al. Generalizing from a few examples: A survey on few-shot learning. ACM Computing Surveys, v. 53, n. 3, art. 63, p. 1–34, 2020. DOI: https://doi.org/10.1145/3386252.
WOLF, Thomas et al. Transformers: State-of-the-art natural language processing. In: Proceedings of the 2020 conference on empirical methods in natural language processing: system demonstrations. 2020. p. 38-45.
YAN, R.; HILL, S. Harnessing artificial intelligence for enhanced veterinary diagnostics. Veterinary Clinical Pathology, v. 53, n. 1, p. 45–56, 2024. DOI: https://doi.org/10.1111/vcp.1340.
ZEILER, M. D.; FERGUS, R. Visualizing and understanding convolutional networks. In: European Conference on Computer Vision (ECCV). Springer, 2014. p. 818–833. DOI: https://doi.org/10.1007/978-3-319-10590-1_53.
Downloads
Publicado
Como Citar
Edição
Seção
Licença
Copyright (c) 2026 Scientific Electronic Archives
Este trabalho está licenciado sob uma licença Creative Commons Attribution 4.0 International License.
A revista se reserva o direito de fazer altera
