Developing a Medical Image Viewer in ClojureScript
summary
Builds a browser DICOM/mammography viewer in ClojureScript that renders 12 megapixel 16-bit images and targets 30 fps; initial prototype read entire images into immutable structures and GC kills performance, so the implementation switches to typed arrays, replaces do-seq with an explicit loop that compiles to fast JS and adds a precomputed lookup table for fast 16→8-bit mapping. Uses clojure.spec for input validation, ships to production, collects thousands of labeled polygon annotations and reports fewer than ten invalid entries, ~200 annotations/day.
AI-written summary · about
chapters (31)
- 0:00Introduction
- 0:13The Kenosha Conspiracy
- 0:56Olivers Background
- 2:59Mammography
- 3:25Problems with Mammography
- 4:46Data Sources
- 5:54What we need
- 6:08Web app
- 7:04Why ClojureScript
- 8:23Redax Documentation
- 9:00WECS Workflow
- 11:13The good news
- 11:42Getting stuck
- 13:34Image Optimization
- 14:44Arrays
- 15:46Performance
- 18:05Lookup Table
- 18:34Computation
- 19:21Annotation
- 22:02Selection Fields
- 22:37Human readable strings
- 22:49Error message
- 23:00Iterating
- 23:24Shipping
- 24:11ClojureScript Experience
- 24:52Offline First Coding
- 26:09Hiring
- 27:57Developer skill spectrum
- 30:13Python startups
- 30:41Django backend
- 33:24Conclusion
related talks
28mHTML Canvas 어디까지 써봤니
31mScientific Computing with Clojure
38mScientific Clojure, a bird's eye view
38m"Growing Data Center networking mgmt UI using ClojureScript, Reagent and re-frame"
36mOptimizing ClojureScript Apps For Speed
27mUnbundling the electronic health record using Clojure
35mClojureScript & User Interfaces: Simplicity yields Possibilities
40mFunctional Client-Side ClojureScript UI Programming with Webfui