Computer Vision Solution for Wound App

The Overview

The Swift Skin and Wound mobile app brings AI to the bedside to help patients and clinicians monitor wound health. Swift partnered with Sama to cost-effectively scale annotations for their CV model, without compromising the clinician-level accuracy they needed.

“We used other annotation providers in the past, but our experience with Sama was superior due to the direct communication we had with the labeling team, who were with us throughout the project.”

Dhanesh Ramachandram, Senior Machine Learning Researcher at Swift Medical

The Challenge

The Swift AutoTissue CV model determines tissue types present within a wound. To provide wound measurement and diagnostics that rivaled or even surpassed the accuracy of trained physicians, Swift Medical needed high-quality annotation at scale.

In-house labeling was too expensive to scale, and outsourced labeling partners could not meet quality requirements. Swift Medical needed a labeling partner who could be upskilled to accurately label a variety of wound tissue types and bounds with clinician-level precision.

97%

Average internal Quality Score

230%

Increase in throughput (week 1 vs week 10)

66.7%

Reduction in rejections (week 1 vs week 10)

The Solution

Swift needed images segmented by different wound area regions and labels for wound type. To ensure clinician-level quality, Sama introduced:

Specialized, in-depth annotator training
Bespoke training from Swift’s clinical team to upskill Sama annotators. 30% subset of dataset labeled, validated, before scaling to the full dataset.

Continuous feedback loops
Slack, email, and phone to align on edge cases and iterate on instructions, resulting in decreased time per task and increased task complexity in the workflow.

Rigorous quality control
Multi-step QA process: AutoQA, dedicated QA team, with Swift sampling weekly and providing feedback to continuously improve quality and throughput.

“Sama gave us visibility into the data labeling process, with tight QA feedback loops to ensure the high standard of quality we required for our models.”

Dhanesh Ramachandram, Senior Machine Learning Researcher at Swift Medical

Conclusion

Expert training, constant communication, and rigorous quality assurance combined with an ML-powered platform enabled the Sama team to deliver a dataset with the high degree of precision that Swift needed: clinician-level accuracy for highly nuanced workflows at scale.

This dataset now powers the Swift mobile app, which autonomously and accurately identifies wound tissue types and proportions. This accurate information is empowering clinicians to make data-backed treatment decisions and create more personalized care plans, improving the overall quality of care they can deliver to their patients.