2023 Fiscal Year Annual Research Report
2D-3D Reconstruction for internal organs using Deep Learning Techniques
| Project/Area Number |
20K20167
|
|---|
| Research Institution | Osaka University |
|---|
Principal Investigator |
武 淑瓊 大阪大学, 産業科学研究所, 助教 (30775763)
|
|---|
| Project Period (FY) |
2020-04-01 – 2024-03-31
|
| Keywords | 2D-3D reconstruction / Deep learning / GAN / CT reconstrcution / X2CTGAN |
|---|
| Outline of Annual Research Achievements |
The objective of this research is to develop algorithms for 2D-3D reconstruction. We pursued this research in two directions. The first direction is based on a pretrained model called X2CTGAN, which reconstructs 3D CT scans from biplanar 2D X-ray images. It achieved a performance score of 0.62 in terms of SSIM. However, the reconstruction quality was far from that required in clinical settings because of a lack of 3D topology information. To solve the issue, we devised a novel approach that utilizes rotations and concatenations of X-ray images. Our new method has demonstrated an almost 30% enhancement in SSIM compared to X2CTGAN. The second direction is sparse-to-dense-view 3D reconstruction. We summarized all these experimental results and planed to publish on international conferences.
|
