AIDeveloper

AIDeveloper trains and evaluates deep neural networks (NN) for image classification, supporting convolutional neural networks (CNN) and other architectures for biological and clinical image analysis.

Key Features:

• Multi-architecture support: Supports convolutional neural networks (CNN) and other neural network architectures for image classification tasks.
• Model training: Trains neural networks on labeled image datasets.
• Performance evaluation: Provides metrics for assessing classification performance.
• Inference on new data: Applies trained models to new datasets for prediction.
• Model exportation: Exports trained models to various formats for use across platforms.

Scientific Applications:

• Object classification (CIFAR-10): Demonstrated by training a convolutional neural network (CNN) on the CIFAR-10 dataset of object images.
• Stem cell differentiation: Distinguishes differentiated versus non-differentiated mesenchymal stem cells (MSCs) in culture.
• Whole blood cell counts: Performs whole blood cell counting using a dataset from real-time deformability cytometry with NN-trained models producing results comparable to conventional clinical methods.
• Label-free immune cell classification: Classifies B- and T-cells derived from human blood without fluorescent labeling.

Methodology:

Training of neural networks (including CNNs) on labeled image datasets, evaluation using classification performance metrics, and export of trained models to multiple formats.