Next generation research applications for hybrid PET/MR and PET/CT imaging using deep learning

Area of Science:

• Medical Imaging
• Artificial Intelligence
• Nuclear Medicine

Background:

• Deep learning and AI, particularly computer vision, show significant promise in medical imaging.
• Convolutional neural networks are increasingly impacting radiology and nuclear medicine.
• AI applications extend beyond image classification to direct image quality enhancement.

Purpose of the Study:

• To review AI applications for improving positron emission tomography (PET) image quality.
• To explore the benefits of integrating anatomical information from CT and MRI.
• To discuss quality assessment and potential biases in AI models.

Main Methods:

• Review of AI techniques applied to PET imaging.
• Integration of AI with hybrid imaging modalities (PET/CT, PET/MRI).
• Discussion on assessing image quality and dataset bias.

Main Results:

• AI can improve PET image quality, enabling shorter scan times and lower radiotracer doses.
• Enhanced image quality leads to improved patient comfort and reduced motion artifacts.
• AI facilitates advanced applications like dual tracer studies and new imaging indications.

Conclusions:

• AI integration in hybrid PET imaging offers significant advancements for research and clinical practice.
• These technologies maximize information extraction from hybrid PET scans.
• Benefits include improved patient care and expanded diagnostic capabilities.