KAUST researchers introduce deepBlastoid, an AI platform automating and standardizing classification of human blastoids from brightfield images with high accuracy and efficiency.
Key Details
- 1deepBlastoid uses a ResNet-18 architecture to classify human blastoid images in five categories.
- 2The curated dataset includes 17,133 brightfield images, with 2,407 expert-labeled.
- 3Achieved up to 97% accuracy by combining the base model (87%) with a Confidence Rate metric for expert fallback.
- 4Processes 273.6 images per second—1,000 times faster than human experts.
- 5Tool validated in drug dose optimization and safety assessment use cases.
- 6Publicly available tools and dataset support global research customization.
Why It Matters
DeepBlastoid exemplifies the potential of AI-powered image analysis for scalable, objective, and reproducible evaluation in developmental biology—a workflow parallel to emerging radiology and digital pathology automation. Its open-source tools may inspire similar strategies for image-intensive tasks in radiology and pathology.
Source
EurekAlert
Related News
•EurekAlert
AI-Powered OCT Enables Rapid 'Optical Biopsy' for Early Endometrial Cancer Detection
A team at Washington University has developed a catheter-based 3D OCT system with AI to quickly and noninvasively detect early endometrial cancers.
•EurekAlert
AI Clinical Reasoning in Diagnostics and Digital Fatigue in Healthcare
Recent JMIR features explore large language models in clinical diagnostics and digital fatigue among healthcare professionals.
•EurekAlert
KAIST, MIT, Microsoft Develop Efficient AI Image Upsampling for Robotics
KAIST, MIT, and Microsoft have created 'Upsample Anything,' a training-free AI method to restore high-resolution visual data from compressed images with up to 16x improved GPU memory efficiency.