Radiologists: Medical Image Review for AI Benchmarks

What We're Researching

We're running a paid study on the accuracy and reliability of diagnostic imaging algorithms. High-quality reference evaluations are essential for establishing safe and effective medical benchmarks. Your clinical expertise will directly inform the baseline ground truth used to evaluate these emerging tools.

How It Works

You will review a series of de-identified medical images and associated clinical cases through a secure remote platform. For each case, you will provide an expert read and complete a structured scoring rubric. The workflow involves assessing specific diagnostic criteria and highlighting key anatomical or pathological findings. All tasks are completed asynchronously, allowing you to evaluate the cases at your own pace.

Who This Is For

We are looking for licensed and practicing radiologists with strong clinical diagnostic experience. We welcome general radiologists, neuroradiologists, musculoskeletal radiologists, and other imaging subspecialists. Candidates must be comfortable applying standardized rubrics to complex diagnostic cases.

What You'll Do

• Review de-identified medical images across various clinical modalities

• Provide expert clinical reads and diagnostic assessments for each case

• Complete structured scoring rubrics based on your findings

• Establish reference evaluations to serve as ground truth for algorithmic benchmarks

Who Should Apply

• Active practice as a radiologist or specialized imaging physician

• Extensive experience interpreting complex clinical cases

• Comfortable providing structured, rubric-based diagnostic evaluations

Compensation

$210 per task

 

Ready to participate?

Start your paid interview now

 

About Terac

Terac is building the world's largest pool of vetted human experts for AI. Researchers, AI labs, and product teams use Terac to recruit, screen, and pay study participants across industries, languages, and skill sets.

 

Learn more at terac.com or on YouTube at @jointerac.