Real-time axial motion detection and correction for single photon emission computed tomography using a linear prediction filter.

Purpose: We have developed an algorithm for real-time detection and complete correction of the patient motion effects during single photon emission computed tomography. The algorithm is based on a linear prediction filter (LPC).

Materials And Methods: The new prediction of projection data algorithm (PPDA) detects most motions-such as those of the head, legs, and hands-using comparison of the predicted and measured frame data. When the data acquisition for a specific frame is completed, the accuracy of the acquired data is evaluated by the PPDA. If patient motion is detected, the scanning procedure is stopped. After the patient rests in his or her true position, data acquisition is repeated only for the corrupted frame and the scanning procedure is continued.

Results: Various experimental data were used to validate the motion detection algorithm; on the whole, the proposed method was tested with approximately 100 test cases. The PPDA shows promising results.

Conclusion: Using the PPDA enables us to prevent the scanner from collecting disturbed data during the scan and replaces them with motion-free data by real-time rescanning for the corrupted frames. As a result, the effects of patient motion is corrected in real time.