紀要論文

深層学習を用いたマルチピンホールSPECT投影データのオーバーラップの除去

REMOVAL OF OVERLAPING IN MULTI-PINHOLE SPECT PROJECTION DATA WITH DEEP LEARNING

jpn

24368083

https://doi.org/10.15002/00030760

法政大学大学院理工学研究科

法政大学大学院紀要. 理工学研究科編

65

1

4

2024-03-24

Version of Record

We propose a separation method of an overlapped projection image in a multi-pinhole single-photon emission computed tomography (SPECT) system. Generally, data acquisition in SPECT systems is performed using a rotating gamma camera. Therefore, data acquisition is time consuming and dynamic study of organs is not possible. In addition, the patient must maintain a certain posture during data acquisition, which is a significant burden. On the other hand, systems using multi-pinhole collimators allow stationary data acquisition in multi-head SPECT systems, which has many advantages such as dynamic data acquisition, reduced patient stress, and shorter data acquisition time. However, one of the drawbacks of this system is its low sensitivity. When the number of pinholes is increased to improve the data acquisition angle and sensitivity, the projected images of the pinholes overlap, causing artifacts in a reconstructed image. To address this problem, avoiding overlapping projection images reduces the spatial resolution of the reconstructed image. In this paper, we propose a method to separate overlapping projection images using CNN in a multi-pinhole SPECT system. As a result of simulations, we confirmed that the overlap of projection images can be removed and converged at an earlier stage than separating them while reconstructing.

Article