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

Two-stage Pipeline for Automated Cell Segmentation: Integrating Semantic and Instance Learning

Thanh-Ha Do, Posts and Telecommunications Institute of Technology, Vietnam, dothanhha@ptit.edu.vn , Hoang Minh-Huong Dang, VNU University of Science, Vietnam, Thanh-Lam Tran, VNU University of Science, Vietnam, Van-De Nguyen, The 108 Military Central Hospital, Vietnam
 
Suggested Citation
Thanh-Ha Do, Hoang Minh-Huong Dang, Thanh-Lam Tran and Van-De Nguyen (2025), "Two-stage Pipeline for Automated Cell Segmentation: Integrating Semantic and Instance Learning", APSIPA Transactions on Signal and Information Processing: Vol. 14: No. 1, e9. http://dx.doi.org/10.1561/116.20250033

Publication Date: 05 Jun 2025
© 2025 T.-H. Do, H. M.-H. Dang, T.-L. Tran and V.-D. Nguyen
 
Subjects
Image and video processing,  Segmentation and grouping,  Color processing,  Deep learning,  Evaluation
 
Keywords
Diff-quick cell imagesmicroscopiccell segmentationautomaticdeep learning
 

Share

Open Access

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

Downloaded: 26 times

In this article:
Introduction 
Deep Neural Networks for Cell Segmentation in Microscopic Imaging 
Single Model vs. Combined Pipeline: Proposed Deep Learning Strategies for Cell Segmentation 
Self-constructed Cell Segmentation Dataset 
Experiment Result 
Conclusion 
References 

Abstract

Accurate segmentation of individual thyroid cells is a prerequisite for cell feature analysis and reliable cancer staging classification. However, Diff-Quick stained cytology images present significant challenges: frequent misclassification of malignant cells and erythrocytes and substantial cell overlap hindering boundary detection. To address these issues, we propose a novel two-stage pipeline. This approach enhanced the efficiency-optimized nnU-Net v2 for rapid foreground-background separation, enabling efficient instance segmentation of overlapping cells and reducing erythrocyte misclassification. The results evaluated on a dataset with multiple thyroid cancer stages show that our method reduced erythrocyte false positives and improved accuracy over the best post-processed baseline while cutting inference time. These findings demonstrate the practical utility of our pipeline for automated Diff-Quick thyroid cytology image segmentation within real-world clinical workflows.

DOI:10.1561/116.20250033