7 Adding AI to Your Hardware
An Introduction to Smart Microscopy
Wesley Legant 
Suliana Manley
- Introduction
- Define and motivate the “problem”
- What is a Biological “Event”?
- Why is it important to study these?
- How is it distinct from an “Object”?
- What is a Biological “Event”?
- Define and motivate the “problem”
- Setting the bounds
- What kinds of signals are we looking to extract events from?
- What spatiotemporal scales are relevant?
- What types of microscopes/imaging assays are in the scope of this chapter?
- Limiting the scope
- Brief history/evolution of automated event/object detection in general
- Offer some background to current methods
- Including classical up to SVM
- e.g micropilot and limitations
- Case Study #1 : CellProfiler
- Case Study #2 : Micropilot
- Including classical up to SVM
- Offer some background to current methods
- The need for new methods
- Deep learning based and ML approaches
- Where the algorithm learns what is important
- What are the advantages of these methods?
- What has been done recently in the literature?
- What is required to label, train and implement these methods?
- Special considerations for event detection modesl
- Specifics of event detection in contast the other ML tasks
- Deep learning based and ML approaches
- Real-time vs a posteriori inference
- Challenges, opportunities
- Event detection as the first step in the microscopy workflow
- Limitations and notes of caution related to inference
- “Trusting the algorithms”
- Conclusion