Pre-conference workshop: Aiforia


12th Jun 2019


New York City, USA
Stewart Hotel

Demystifying Deep Learning AI and Digital Pathology:

Demystifying Deep Learning AI and Digital Pathology:


Your one-stop workshop will help you to understand and deploy intelligent solutions with ‘Aiforia Create’ Deep Learning Pathology. With a hands on experience to ensure you are trained to achieve precise active annotations and understand how different annotation mistakes affect your results.

The hands-on part is run as a friendly competition and the lowest error rate walks away with a prize and bragging rights.


  • A road map for implementation of Deep Learning solutions for your Digital Pathology workflow.
  • A deep understanding for the importance of ‘high quality, not high quantity’ of training data.
  • Understand how to structure algorithms,
  • Train your own Deep Learning Neural Network, Live, On the Premises.

Join the discussion and share your thoughts and ideas with us and the invited guest speakers


Thomas Westerling-Bui, Ph.D. , Senior Scientist, Aiforia Inc

Digital Pathology for drug development in the age of unlimited analysis design possibilities

Automated analysis can improve precision and recall, as well as reduce inter- and intra-operator differences. This leads to increased effect size and likelihood to get a measure of the true effect size of an experimental drug treatment.

Gillian Beamer, Assistant Professor, Tufts University

End user developed Deep Learning algorithms for the analysis of Lung granulomas in experimental tuberculosis.

Exploring the utility of domain expert driven Deep Learning algorithms to study Host Response to virulent M. tuberculosis including algorithms to delineate and segment granuloma and necrosis in hematoxylin and eosin (H&E) stained lung tissue from M. tuberculosis-infected Diversity Outbred mice.

  • Examples of user generated Deep Learning algorithms in pre-clinical research models.
  • Paths to algorithm validation
  • Path to deploying Create for yourself or your team

Peter Westcott, Postdoctoral fellow, Koch Institute for Integrative Cancer Research

Deep neural network for automatic histopathological analysis and grading of murine lung tumors

A demonstration that deep neural networks can be used for automated analysis and grading of preclinical models of lung cancer and an exploration of how this powerful technology has the potential to increase the throughput, sensitivity and reproducibility of hypothesis-driven studies of factors influencing tumor progression and immune response in mouse models of lung cancer.

  • What happens when an entire lab at MIT uses one unifying digital pathology algorithm instead of multiple individual human operators.
  • Development of a tumor grading algorithm that is robust against known usual tissue artefacts.
  • Paths to Algorithm Validation

Jean-Martin Lapointe, DVM, MSc, DACVP, Senior Pathologist, AstraZeneca, UK

Deep learning image analysis approaches for evaluation of NAFLD/NASH mouse model histopathology

Nonalcoholic fatty liver disease/steatohepatitis (NAFLD/NASH) are common human liver diseases. Subjective scoring systems devised to diagnose histopathologic features of NAFLD/NASH have been adapted to animal models, but such semi-quantitative data shows poor granularity and statistical applicability, so has limited research usefulness. Image analysis is crucial to obtain quantitative data, but has its own challenges in liver pathology. H&E-stained liver will show hepatocyte lipid as clear spaces, difficult to distinguish from processing-associated glycogen loss. Total collagen can be quantified by IHC, but fibrosis-associated and normal structural collagen cannot be easily differentiated. Biliary hyperplasia does not show sufficiently distinctive H&E features for image analysis detection. AI deep learning allows for discriminating features of interest in H&E or IHC stains. Algorithms were developed to quantify macrovesicular and microvesicular steatosis, lobular vs portal collagen, and bile ducts, providing the quantified, granular measurements necessary to evaluate NAFLD/NASH in animal models


Register for the Workshop Here





Stewart Hotel

371 7th Ave, New York, NY 10001, USA

Return to the Main Conference