Systems and methods for cardiac triggering of an imaging system

Patent metadata

Field	Value
Publication number	US-11607181-B2
Application number	US-201816041462-A
Country	US
Kind code	B2
Filing date	Jul 20, 2018
Priority date	Jul 20, 2018
Publication date	Mar 21, 2023
Grant date	Mar 21, 2023

Abstract

Methods and systems are provided for cardiac triggering of an imaging system. a method for an imaging system comprises acquiring, during a scan of a subject, an electrical signal indicating a periodic physiological motion of an organ of the subject, inputting a sample of the electrical signal into a trained neural network to detect whether a peak is present in the sample, triggering acquisition of image data responsive to detecting the peak in the sample, and not triggering the acquisition of image data responsive to not detecting the peak in the sample. In this way, the timing of data acquisition may be optimally and robustly synchronized with a cardiac cycle.

First claim

The invention claimed is:

1. A method for an imaging system, comprising:
acquiring, during a scan of a subject, an electrical signal indicating a periodic physiological motion of an organ of the subject;
pre-processing the electrical signal to generate a plurality of samples,
wherein the plurality of samples comprise a first sample and a second sample, and
wherein the first sample and the second sample comprises a plurality of discrete measurements of the electrical signal over time during the scan;
inputting the first sample of the electrical signal into a trained neural network to detect whether a peak is present in the first sample;
triggering acquisition of image data responsive to detecting the peak in the first sample;
not triggering the acquisition of image data responsive to not detecting the peak in the first sample; and
responsive to not detecting the peak in the first sample, discarding the first sample, inputting the second sample of the electrical signal into the trained neural network, and triggering the acquisition of image data responsive to detecting the peak in the second sample,
wherein the second sample at least partially temporally overlaps the first sample in the electrical signal.

2. The method of claim 1 , wherein pre-processing the electrical signal further comprises converting the electrical signal from digital values to millivolts.

3. The method of claim 1 , wherein the plurality of discrete measurements of the electrical signal over time during the scan comprises a raw sample, and wherein pre-processing the electrical signal further comprises applying one or more bandpass filters to the raw sample to generate one or more filtered samples, wherein the first sample comprises the raw sample and the one or more filtered samples.

4. The method of claim 1 , further comprising acquiring a reference signal indicating the periodic physiological motion of the organ of the subject prior to the scan of the subject to generate a reference sample, and inputting the reference sample with the first sample to the trained neural network.

5. The method of claim 4 , wherein the reference signal comprises electrocardiograph (ECG) data of the subject when ECG sensors were not subjected to magnetic fields and radio frequency (RF) signals.

6. The method of claim 1 , further comprising inputting the first sample into a signal quality classifier to determine whether a signal quality of the first sample is degraded, and rejecting output of the trained neural network responsive to the signal quality of the first sample being degraded.

7. The method of claim 1 , wherein the organ comprises a heart, and wherein the electrical signal comprises an electrocardiogram.

8. The method of claim 7 , wherein the trained neural network comprises a convolutional neural network, and wherein the peak comprises an R-peak of the electrocardiogram.

9. The method of claim 1 , wherein the trained neural network includes three output nodes: peak presence, no peak presence, and non-peak spike noise.

10. A method, comprising:
acquiring, during a diagnostic scan of a subject, a first channel and a second channel of an electrical signal indicating periodic physiological motion of an organ of the subject;
pre-processing the first channel and the second channel of the electrical signal into a first sample and a second sample, respectively,
wherein the first sample and the second sample comprises a plurality of discrete measurements of the electrical signal over time during the diagnostic scan;
inputting the first sample into a first trained neural network to detect whether a peak in the periodic physiological motion is present in the first sample;
inputting the second sample into a second trained neural network to detect whether the peak is present in the second sample,
wherein the first channel is different from the second channel, and
wherein the second sample at least partially temporally overlaps the first sample in the electrical signal;
determining presence of the peak by combining detections of the first trained neural network and the second trained neural network; and
triggering acquisition of image data responsive to determining that the peak is present.

11. The method of claim 10 , further comprising: inputting the first sample into a first signal quality classifier to determine the signal quality of the first sample; inputting the second sample into a second signal quality classifier to determine the signal quality of the second sample; and determining presence of the peak by combining detections of the first trained neural network and the second trained neural network, and the signal quality of the first sample and the second sample.

12. The method of claim 11 , further comprising:
responsive to the signal quality of the first sample and the signal quality of the second sample both being not degraded, determining a timing of triggering according to a phase-corrected average of output from a first peak detection classifier and output from a second peak detection classifier;
responsive to the signal quality of the first sample being not degraded and the signal quality of the second sample being degraded, determining the timing according to the output of the first peak detection classifier; and
responsive to the signal quality of the first sample being degraded and the signal quality of the second sample being not degraded, determining the timing according to the output of the second peak detection classifier.

13. The method of claim 12 , further comprising, responsive to the signal quality of the first sample being degraded and the signal quality of the second sample being degraded, not triggering the acquisition of image data, and outputting an indication that the acquisition of the electrical signal is degraded.

14. A system, comprising:
a medical scanning system for scanning a subject;
an electrocardiograph (ECG) sensor positioned on the subject for generating an ECG signal; and
a processor communicatively coupled to the medical scanning system and the ECG sensor and configured to:
acquire a reference ECG signal from the subject via the ECG sensor prior to a scan of the subject to generate a reference sample;
acquire, during the scan, the ECG signal via the ECG sensor;
pre-processing the ECG signal into a sample that comprises a plurality of discrete measurements of the ECG signal over time during the scan;
input both the sample of the ECG signal and the reference sample into a trained neural network to detect whether a peak is present in the sample;
trigger acquisition of image data by the medical scanning system responsive to detecting the peak in the sample; and
not trigger the acquisition of image data by the medical scanning system responsive to not detecting the peak in the sample.

15. The system of claim 14 , wherein the plurality of discrete measurements of the ECG signal over time during the scan comprises a raw sample, and wherein pre-processing the ECG signal further comprises applying one or more passband filters to the raw sample to generate one or more filtered samples, wherein the sample comprises the raw sample and the one or more filtered samples.

Assignees

• Gen Electric
• Ge Prec Healthcare Llc

Inventors

• Sha Liewei
• Gui Dawei
• Liu Yawei
• Ma Yan
• Wang Haonan

Classifications

• A61B5/33

  specially adapted for cooperation with other devices · CPC title

• A61B5/7285 Primary

  for synchronizing or triggering a physiological measurement or image acquisition with a physiological event or waveform, e.g. an ECG signal · CPC title

• A61B5/055 Primary

  involving electronic [EMR] or nuclear [NMR] magnetic resonance, e.g. magnetic resonance imaging · CPC title

• A61B5/7264

  Classification of physiological signals or data, e.g. using neural networks, statistical classifiers, expert systems or fuzzy systems · CPC title

• A61B5/0044

  for the heart · CPC title