1. DeepQ AI Platform overview

DeepQ AI Platform provides a one-stop solution for your medical image MLOPs cycle.

• DeepQ AI Platform dramatically reduces the learning thresholds and the cost of AI model training through efficient image annotation tools, optimized training environment, built-in multiple AI models, fully automated parameter tuning, and a simple user interface.

• It is different from the existing AI training platforms and services available on the market, users of DeepQ AI Platform can quickly convert their own medical image data into the AI models without the knowledge and skills of deep learning.

The complete DeepQ AI Platform consists of three main modules：

• **DeepCap：**Dataset management & image annotation

• **AI Training：**Model selection/tuning & evaluation

• **Deeploy：**Real-time model deployment, worklist & AI viewer

DeepCap (Annotation tool)

• De-identification & anonymize of the imported medical image data.

• To produce labeled data using annotation tools (classification, detection & segmentation).

• To create, manage & monitor annotation projects and to ensure the quality of labeled data.

• To import labeled data from other sources.

AI training

• pre-built AI models for best training results.

• Create, monitor & manage AI training tasks.

• No-code automatic hyper-parameter tuning that can leverage advanced techniques such as transfer learning & data augmentation.

• Evaluate the training progress & outcome of the models with different metrics.

Deeploy (Model deployment)

• Get your AI models up and running with real-world data sent from PACS

• Inference engine automatically identifies image by DICOM tag and forwards to different AI models

• Set deployment authorization. Different users for different model deployments

• Download inference history for futher analysis and model retrain.