Classifying Medical Imaging On-Device with Edge Impulse and BrainChip
- BrainChip and Edge Impulse collaborate to develop a medical diagnostics system.
https://www.edgeimpulse.com/blog/ai-xray-vision/
Summary of this article:
- BrainChip's Akida neuromorphic processor and Edge Impulse's platform used in the project.
- Proof of concept created for diagnosing pneumonia from X-ray images.
- BrainChip Akida Raspberry Pi Developer Kit employed for the project.
- Training dataset obtained from Kaggle containing nearly 6,000 chest X-ray images.
- Training process resulted in a 97% accuracy rate.- Model tested against separate data for validation, achieving over 94% accuracy.- Model deployed to BrainChip Akida Raspberry Pi Developer Kit.
- Two inference methods demonstrated: using Edge Impulse Linux SDK for image file classification and live stream classification from a USB webcam.
- Inference times ranged from 100 to 150 milliseconds.
- Akida chip demonstrated low power consumption (milliwatts).
- Potential applications extend beyond medical diagnostics.
- Edge Impulse project made public for further exploration and adaptation.
Classifying Medical Imaging On-Device with Edge Impulse and...
Add BRN (ASX) to my watchlist
 (20min delay)
|
|||||
|
Last
25.3¢ |
Change
-0.003(0.98%) |
Mkt cap ! $473.2M | |||
| Open | High | Low | Value | Volume |
| 25.5¢ | 27.0¢ | 25.0¢ | $1.688M | 6.550M |
Buyers (Bids)
| No. | Vol. | Price($) |
|---|---|---|
| 57 | 1546093 | 25.0¢ |
Sellers (Offers)
| Price($) | Vol. | No. |
|---|---|---|
| 25.5¢ | 74985 | 7 |
View Market Depth
| Last trade - 15.09pm 22/05/2024 (20 minute delay) ? |
|
|||||
|
Last
25.5¢ |
Change
-0.003 ( 0.23 %) |
||||
| Open | High | Low | Volume | ||
| 25.5¢ | 26.5¢ | 25.0¢ | 4381415 | ||
| Last updated 15.31pm 22/05/2024 ? | |||||
| BRN (ASX) Chart |