Development of algorithms to automate the detection of lesions

Development of algorithms to automate the detection of lesions

The Customer:

One of the leading medical imaging device manufacturers from Japan

Project Description & Scope:

• Image processing algorithm development for automated defect detection in the retina
• Detection of Dark & Bright Lesions

The Solution:

• Identified algorithms for Ganglion Cell layer thickness detection
• Refinement of algorithm for target images and performance fine-tuning
• Image processing solutions: Image segmentation, preprocessing & post processing
  • Pre-processing: Detection of optical disk, blood vessel structure and fovea
  • Post-processing: Rejection of false detection based on pre-defined constraints

Business Benefit:

• Experienced team in the medical imaging domain improving overall product features by taking complete ownership of application development
• Completed development in lesser time against target time thus reducing cost & faster Time-to-Market for product launch
• Leveraged Tata Elxsi medical device in-house regulatory compliance practice for FDA compliance
• A fixed-cost engagement in the rapid-product prototyping development helped the customer keep risks under control