Back to all blog posts
Introduction
Recent advancements in technology have brought together artificial intelligence (AI) and veterinary medicine. As AI models have become accessible through user-friendly platforms, they are now increasingly being utilized by pathologists in both preclinical research and clinical assessment. Veterinary pathologists study disease in animals. They serve as some of the leading experts in preclinical drug development, as well as in diverse research areas where the use of animal models prevails.
The study of disease is a precise craft, and the tools traditionally used in the discipline require time and attention in order to achieve consistent results. Pathologists often need to examine microscopic structures and phenomena, working out cell counts, area measurements, and distance calculations manually. This type of work is often time-consuming and prone to bias. The analysis is based on visual interpretation, and thus dependent on subjective judgment. Also, fatigue brought by manual work can affect the accuracy of results.
How does artificial intelligence fit into the picture of veterinary pathology? As in many fields, emerging technologies are used to automate tasks that are tedious but vital for accurate results and confident decision-making. In pathology, AI can serve as an image analysis tool to aid the expert in their work. The neural networks of deep learning AI models can today be taught to identify, quantify, measure, and gather other data from microscopic images. This frees up valuable resources, as a big part of analyzing samples can now reliably be computerized. Software platforms allow pathologists to effectively examine microscopic slides with the help of Whole Slide Imaging (WSI), and ready-to-use AI tools for image analysis.
How do AI applications apply in the real world? To clarify, we have highlighted three use cases from research and diagnostics where deep learning AI models are used by pathologists.
Research: Brain areas and cell types
AI-powered image analysis is used in cellular neuroscience to analyze sample tissues on microscopic slides. The technology is utilized to quantify brain areas, and counting specific cell types within them. The AI model is able to calculate thousands of cells in less than a minute. Finally, distance calculations between cells and area borders can be extracted from the images with the help of the analytical tools.
Diagnostics: Cutaneous mast cell tumor Ki67
Ki67 stainings and the Ki67-index is used to predict the survival in cutaneous mast cell tumor canines. As the approach serves as a good diagnostics tool, counting the stained cells is a time-consuming task. However, reliable results can be achieved with automated image analysis, as the AI-powered tools do the job in just some seconds, giving veterinary pathologists a needed leverage to their diagnostic work.
Research: Screening of bone marrow cellularity changes
Bone marrow is a vital tissue for analysis in pharmacologic safety studies. The cellularity assessment traditionally consists of cell counts and applied mathematical modeling, a time-consuming and non-practical task at scale. AI models can be trained to differentiate hematopoietic cells from the tissue. This automates cell counting and provides an objective and a replicable measure of the bone marrow cellularity. Read more about Veterinary Pathologist Mark Smith's work on AI-assisted screening of bone marrow cellularity changes.
Technological advancements are helping pathologists reach more accuracy with less effort. AI-powered image analysis is gaining popularity among experts and gaining a place in veterinary medicine as a standard, under-the-belt tool. As more segments of the work are automated, further resources can be used for critical tasks that require higher problem-solving skills and creativity.
See the above case studies in action on Aiforia’s AI for image analysis software.