Sign determination methods for the respiratory signal in data-driven PET gating

Area of Science:

• Medical Imaging
• Nuclear Medicine
• Computational Imaging

Background:

• Patient respiratory motion during PET scans causes image blurring and artifacts, reducing lesion detectability and treatment planning accuracy.
• Current respiratory gating relies on external devices, which can be uncomfortable and require setup.
• Data-driven (DD) methods offer motion signals directly from PET data, eliminating external equipment but often lack directional information.

Purpose of the Study:

• To develop novel methods (CorrWeights and CorrSino) for determining the correct direction of respiratory motion signals derived from PET data.
• To evaluate the accuracy and efficiency of these new methods compared to existing registration-based approaches.
• To enable more accurate motion correction in PET imaging without external devices.

Main Methods:

• Proposed two novel methods, CorrWeights and CorrSino, utilizing PET raw data to determine the direction of data-driven respiratory signals.
• Applied principal component analysis (PCA) to acquired PET data to generate motion signals.
• Implemented and compared with two versions of a published registration-based method requiring image reconstruction.
• Evaluated methods on XCAT simulations and real cancer patient datasets using Varian Real-time Position Management (RPM) device.

Main Results:

• CorrWeights and CorrSino accurately detected the correct direction of respiratory motion signals.
• Novel methods demonstrated lower failure rates (<3%) compared to registration-based methods when DD signals correlated with RPM.
• The proposed methods showed faster computation times and avoided the need for image reconstruction.
• CorrWeights is versatile and can be integrated with various DD methods.

Conclusions:

• CorrWeights and CorrSino effectively resolve the directional ambiguity in data-driven respiratory signals for PET imaging.
• These methods enhance the accuracy of motion correction, potentially improving lesion detection and treatment planning.
• The developed techniques offer a more efficient and patient-friendly alternative to external respiratory gating systems.