APSIPA Transactions on Signal and Information Processing > Vol 14 > Issue 1

PSHop: A Lightweight Feed-Forward Method for 3D Prostate Gland Segmentation

Yijing Yang, University of Southern California, USA, Vasileios Magoulianitis, University of Southern California, USA, magoulia@usc.edu , Jiaxin Yang, University of Southern California, USA, Jintang Xue, University of Southern California, USA, Masatomo Kaneko, University of Southern California, USA, Giovanni Cacciamani, University of Southern California, USA, Andre Abreu, University of Southern California, USA, Vinay Duddalwar, University of Southern California, USA, C.-C. Jay Kuo, University of Southern California, USA, Inderbir S. Gill, University of Southern California, USA, Chrysostomos Nikias, University of Southern California, USA
 
Suggested Citation
Yijing Yang, Vasileios Magoulianitis, Jiaxin Yang, Jintang Xue, Masatomo Kaneko, Giovanni Cacciamani, Andre Abreu, Vinay Duddalwar, C.-C. Jay Kuo, Inderbir S. Gill and Chrysostomos Nikias (2025), "PSHop: A Lightweight Feed-Forward Method for 3D Prostate Gland Segmentation", APSIPA Transactions on Signal and Information Processing: Vol. 14: No. 1, e6. http://dx.doi.org/10.1561/116.20240061

Publication Date: 30 Apr 2025
© 2025 Y. Yang, V. Magoulianitis, J. Yang, J. Xue, M. Kaneko, G. Cacciamani, A. Abreu, V. Duddalwar, C. C. J. Kuo, I. S. Gill and C. Nikias
 
Subjects
Pattern recognition and learning,  Biological and biomedical signal processing,  Image and video processing,  Multiresolution signal processing,  Medical image analysis,  Classification and prediction
 
Keywords
Magnetic resonance imagingprostate gland segmentationdata-driven radiomicsfeed-forward modelinterpretable pipeline
 

Share

Open Access

This is published under the terms of CC BY-NC.

Downloaded: 7 times

In this article:
Introduction 
Related Work 
Methods 
Experimental Setup 
Experimental Results 
Conclusion 
References 

Abstract

Automatic prostate segmentation is an important step in computer-aided diagnosis of prostate cancer and treatment planning. Existing methods of prostate segmentation are based on deep learning models which have a large size and lack of transparency which is essential for physicians. In this paper, a new data-driven 3D prostate segmentation method on MRI is proposed, named PSHop. Different from deep learning based methods, the core methodology of PSHop is a feed-forward encoder-decoder system based on successive subspace learning (SSL). It consists of two modules: 1) encoder: fine to coarse unsupervised representation learning with cascaded VoxelHop units, 2) decoder: coarse to fine segmentation prediction with voxel-wise classification and local refinement. Experiments are conducted on the publicly available ISBI-2013 dataset, as well as on a larger private one. Experimental analysis shows that our proposed PSHop is effective, robust and lightweight in the tasks of prostate gland and zonal segmentation, achieving a Dice Similarity Coefficient (DSC) of 0.873 for the gland segmentation task. PSHop achieves a competitive performance comparatively to other deep learning methods, while keeping the model size and inference complexity an order of magnitude smaller.

DOI:10.1561/116.20240061