Automated Multimodal Image Registration for Prostate Cancer Using Squeeze-and-Excitation ResNet with Thin Plate Spline Transformation: A Deep Learning Approach

19 October 2025 : Laboratory Research

Hua Cai^{ABCDEFG 1}, Yongxin Li^{C 2}, Qiang Zhao^{F 3*}, Yun Lu^{A 1,4}

DOI: 10.12659/MSM.949996

Med Sci Monit 2025; 31:e949996

Authors information Article notes Copyright and License information

Introduction Material and Methods Results Discussion Conclusions References

Related articles Order reprints Share article Share by email

View HTML version

Table 1 Comparative registration performance (prostate region) or performance comparison of different methods on prostate registration.

Methods	DSC	HD (mm)	Time (s)
CNNGeometric	0.945 (±0.030)	5.32 (±1.5)	5.2 (±4.0)
VoxelMorph	0.924 (±0.009)	6.72 (±1.6)	12.4 (±3.6)
SE-ResNet-TPS	0.964 (±0.008)	2.91 (±0.5)	6.0 (±2)

Back to the Article