Traditionally, histopathology involves the manual microscopic analysis of tissue mounted on glass slides. Now, the once ubiquitous laboratory microscope is being rapidly replaced by advanced scanning apparatus, which digitises slides for display on PC monitors. However, tissue analysis remains a highly skilled and time-consuming process, requiring years of veterinary and specialist training to achieve proficiency. In veterinary pathology, this training, undertaken by veterinary surgeons, ultimately leads to board certification (e.g., European College of Veterinary Pathologists, American College of Veterinary Pathologists, or Royal College of Pathologists).

The role of artificial intelligence (AI) in our workplaces and daily lives is a ‘hot topic’. AI is increasingly being utilised across various sectors and has the potential to revolutionise both veterinary and human pathology. However, to obtain meaningful data and outcomes, the current technology is heavily dependent on training by expert pathologists. As such, at least for the foreseeable future, AI is not likely to entirely replace pathologists, but has the potential to increase efficiency and consistency – this provides benefit to the sector by helping to close the gap between the increasing demand for pathologists and relatively small number of board-certified pathologists.

Currently, in the pathology speciality, AI is most widely used for image analysis – specifically, interpreting histological images. Using expert pathologist input, artificial neural networks can be trained to recognise specific patterns associated with particular disease processes. For example, well-differentiated adenocarcinomas infiltrating tissue often form repeated epithelial-lined tubular structures. AI can ‘learn’ these patterns and assist pathologists by identifying similar structures in other tissues, allowing for the rapid detection of carcinoma metastases, for example. Additionally, AI could supervise the generation of pathology reports based on key histological features or biochemical data derangements, streamlining diagnoses and suggesting the most likely diagnosis or differential diagnoses to the anatomical pathologists reviewing a case.

AI in Human Pathology

AI has been extensively employed as a diagnostic assistance tool in human pathology. Notable examples include:

• Tumour Detection and Grading: AI improves breast and prostate cancer diagnostics by enhancing accuracy and consistency (Nassif et al., 2022; Goldenberg et al., 2019).
• Prognostic Factor Identification: AI can analyse large datasets to predict patient outcomes, such as in human cancers (Torrente et al., 2022) and COVID-19 progression (Jiao et al., 2021).
• Cell Classification and Counting: AI tools offer faster, more consistent classification and counting of cell types, such as immune cell densities in tumour micro-environments (Rakha et al., 2021).

A relatively recent review article involving medics/histopathologists at the University of Nottingham, UK (Rakha et al., 2021) has identified numerous current and future uses of AI in the field. In short, AI can aid with standardising diagnoses in patients, specifically by allowing diagnoses to be more consistent and reproducible, and with increased speed. This can include streamlining defined points within the histopathology workflow, such as automatically requesting special stains, prioritising certain cases, automatically identifying quality control (QC) issues etc.

AI in Veterinary Pathology

While AI has shown significant yet early promise in human pathology, its use in veterinary pathology remains limited and in its infancy. However, recent AI applications in veterinary and toxicological pathology include:

• Liver Fibrosis Assessment: AI has successfully quantified hepatic fibrosis in preclinical mouse models with accuracy comparable to board-certified toxicological pathologists (Ramot et al., 2021), with the potential of reducing manual workloads.
• Canine and Feline Lymphoma Diagnosis: AI models can predict chemotherapy responses and subtype lymphomas by nuclear sizing, improving diagnostic accuracy (Koo et al., 2021; Haghofer et al., 2023).
• Feline Chronic Enteropathy: AI-based models can quantify lymphocytes in feline intestinal biopsies, increasing reproducibility (Wulcan et al., 2024).
• Bovine Mastitis Detection: AI may assist in early mastitis diagnosis, benefiting dairy industry efficiency and animal welfare (Mitsunaga et al., 2024).

Note that these examples are still in the realms of research based applications and are not currently commercially available tests.

Challenges and Ethical Considerations

Despite its potential, AI integration in pathology does face several challenges. Specific points to carefully consider include:

• Data Quality: AI training relies on vast datasets, potentially containing erroneous data (e.g. human error) that can impact model performance. Furthermore, a solid ‘ground truth’ is required – this is the information that the machine ‘learns’. If a group of pathologists cannot agree on diagnosis (i.e. the diagnosis is ambiguous), then AI cannot be expected to make an accurate diagnosis.
• Decision Making: As highlighted in a recent article (Rakha et al., 2021), neural networks operate as “black boxes,” (i.e. a system coming to a conclusion without a clear explanation for how a decision was made) making it difficult to understand and reproduce their decision-making processes.
• Ethical Concerns: Issues such as data ownership and privacy may need to be addressed before AI models, trained on such data, are widely adopted or integrated into workflows.
• Validation Standards: AI tools must be rigorously validated to ensure performance matches that of board-certified histopathologists. Few standardised guidelines exist for this validation process (Ma et al., 2025).

Conclusion

AI has the potential to transform the pathology speciality by increasing diagnostic efficiency, increasing diagnostic accuracy and reducing the workload for pathologists. However, it is not yet to be considered a replacement for human expertise but rather a tool for diagnostic assistance. As AI continues to evolve, the use of careful validation, consideration of ethical implications, and use of high-quality training datasets will be essential for its successful integration into veterinary pathology.

AI is a rapidly developing field which has much to offer the field of veterinary pathology, and also the services that VPG may offer in the future. As we endeavour to be experts and challengers, AI has the potential to offer more precise, standardised, and efficient diagnostic reporting, leading to more rapid and accurate clinical diagnoses for your clients and their pets. Ultimately this should lead to improved patient care and outcomes.

At the VPG, we have recently introduced AI-assessment of the Ki67 index in canine cutaneous mast cell  tumours. This topic will be discussed in Part 2 of this blog.

References

Goldenberg, S., et al. A new era: artificial intelligence and machine learning in prostate cancer. Nat Rev Urol 2019, 16, 391–403

Haghofer A., et al,. Histological classification of canine and feline lymphoma using a modular approach based on deep learning and advanced image processing. Sci Rep. 2023, 13, 19436

Jiao, Z., et al. Prognostication of patients with COVID-19 using artificial intelligence based on chest x-rays and clinical data: a retrospective study. The Lancet Digital Health, 3, e286 – e294

Koo J., et al. Predicting dynamic clinical outcomes of the chemotherapy for canine lymphoma patients using a machine learning model. Vet Sci. 2021, 8, 301

Ma, Y, et al. AI in Histopathology Explorer for comprehensive analysis of the evolving AI landscape in histopathology. npj Digit. Med. 2025, 8, 156

Mitsunaga TM, et al., Current trends in artificial intelligence and bovine mastitis research: a bibliometric review approach. Animal. 2024, 9, 14

Nassif, A., et al. Breast cancer detection using artificial intelligence techniques: A systematic literature review, Artif Intell Med, 2022, 127, 102276

Rakha, EA., et al. Current and future applications of artificial intelligence in pathology: a clinical perspective. J Clin path. 2021, 74, 409-414

Ramot Y., et al, Microscope-based automated quantification of liver fibrosis in mice using a deep learning algorithm. Toxicol Pathol. 2021, 49, 1126-1133

Torrente, M., et al. An artificial intelligence-based tool for data analysis and prognosis in cancer patients: results from the clarify study. Cancers 2022, 14, 4041

Wulcan JM., et al., Artificial intelligence-based quantification of lymphocytes in feline small intestinal biopsies. Vet Path. 2024, 62, 139-151