A Hybrid Deep Learning Framework for Accurate Cell Segmentation in Whole Slide Images Using YOLOv11, StarDist, and SAM2

Accurate segmentation of cellular structures in whole slide images (WSIs) is essential for quantitative analysis in computational pathology. However, the complexity and scale of WSIs present significant challenges for conventional segmentation methods. In this study, we propose a novel hybrid deep l...

Full description

Saved in:
Bibliographic Details
Main Authors: Julius Bamwenda, Mehmet Siraç Özerdem, Orhan Ayyıldız, Veysı Akpolat
Format: Article
Language:English
Published: MDPI AG 2025-06-01
Series:Bioengineering
Subjects:
Online Access:https://www.mdpi.com/2306-5354/12/6/674
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1849467425227014144
author Julius Bamwenda
Mehmet Siraç Özerdem
Orhan Ayyıldız
Veysı Akpolat
author_facet Julius Bamwenda
Mehmet Siraç Özerdem
Orhan Ayyıldız
Veysı Akpolat
author_sort Julius Bamwenda
collection DOAJ
description Accurate segmentation of cellular structures in whole slide images (WSIs) is essential for quantitative analysis in computational pathology. However, the complexity and scale of WSIs present significant challenges for conventional segmentation methods. In this study, we propose a novel hybrid deep learning framework that integrates three complementary approaches, YOLOv11, StarDist, and Segment Anything Model v2 (SAM2), to achieve robust and precise cell segmentation. The proposed pipeline utilizes YOLOv11 as an object detector to localize regions of interest, generating bounding boxes or preliminary masks that are subsequently used either as prompts to guide SAM2 or to filter segmentation outputs. StarDist is employed to model cell and nuclear boundaries with high geometric precision using star-convex polygon representations, which are particularly effective in densely packed cellular regions. The framework was evaluated on a unique WSI dataset comprising 256 × 256 image tiles annotated with high-resolution cell-level masks. Quantitative evaluations using the Dice coefficient, intersection over union (IoU), F1-score, precision, and recall demonstrated that the proposed method significantly outperformed individual baseline models. The integration of object detection and prompt-based segmentation led to enhanced boundary accuracy, improved localization, and greater robustness across varied tissue types. This work contributes a scalable and modular solution for advancing automated histopathological image analysis.
format Article
id doaj-art-2a23a7118729471fa3ac9f59dd42e7a4
institution Kabale University
issn 2306-5354
language English
publishDate 2025-06-01
publisher MDPI AG
record_format Article
series Bioengineering
spelling doaj-art-2a23a7118729471fa3ac9f59dd42e7a42025-08-20T03:26:15ZengMDPI AGBioengineering2306-53542025-06-0112667410.3390/bioengineering12060674A Hybrid Deep Learning Framework for Accurate Cell Segmentation in Whole Slide Images Using YOLOv11, StarDist, and SAM2Julius Bamwenda0Mehmet Siraç Özerdem1Orhan Ayyıldız2Veysı Akpolat3Engineering Faculty, Electrical & Electronics Engineering Department, Dicle University, 21280 Diyarbakır, TürkiyeEngineering Faculty, Electrical & Electronics Engineering Department, Dicle University, 21280 Diyarbakır, TürkiyeMedical Faculty, Department of Internal Medicine-Hematology, Dicle University, 21280 Diyarbakır, TürkiyeMedical Faculty, Department of Biophysics, Dicle University, 21280 Diyarbakır, TürkiyeAccurate segmentation of cellular structures in whole slide images (WSIs) is essential for quantitative analysis in computational pathology. However, the complexity and scale of WSIs present significant challenges for conventional segmentation methods. In this study, we propose a novel hybrid deep learning framework that integrates three complementary approaches, YOLOv11, StarDist, and Segment Anything Model v2 (SAM2), to achieve robust and precise cell segmentation. The proposed pipeline utilizes YOLOv11 as an object detector to localize regions of interest, generating bounding boxes or preliminary masks that are subsequently used either as prompts to guide SAM2 or to filter segmentation outputs. StarDist is employed to model cell and nuclear boundaries with high geometric precision using star-convex polygon representations, which are particularly effective in densely packed cellular regions. The framework was evaluated on a unique WSI dataset comprising 256 × 256 image tiles annotated with high-resolution cell-level masks. Quantitative evaluations using the Dice coefficient, intersection over union (IoU), F1-score, precision, and recall demonstrated that the proposed method significantly outperformed individual baseline models. The integration of object detection and prompt-based segmentation led to enhanced boundary accuracy, improved localization, and greater robustness across varied tissue types. This work contributes a scalable and modular solution for advancing automated histopathological image analysis.https://www.mdpi.com/2306-5354/12/6/674WSIcell segmentationSAM2deep learning
spellingShingle Julius Bamwenda
Mehmet Siraç Özerdem
Orhan Ayyıldız
Veysı Akpolat
A Hybrid Deep Learning Framework for Accurate Cell Segmentation in Whole Slide Images Using YOLOv11, StarDist, and SAM2
Bioengineering
WSI
cell segmentation
SAM2
deep learning
title A Hybrid Deep Learning Framework for Accurate Cell Segmentation in Whole Slide Images Using YOLOv11, StarDist, and SAM2
title_full A Hybrid Deep Learning Framework for Accurate Cell Segmentation in Whole Slide Images Using YOLOv11, StarDist, and SAM2
title_fullStr A Hybrid Deep Learning Framework for Accurate Cell Segmentation in Whole Slide Images Using YOLOv11, StarDist, and SAM2
title_full_unstemmed A Hybrid Deep Learning Framework for Accurate Cell Segmentation in Whole Slide Images Using YOLOv11, StarDist, and SAM2
title_short A Hybrid Deep Learning Framework for Accurate Cell Segmentation in Whole Slide Images Using YOLOv11, StarDist, and SAM2
title_sort hybrid deep learning framework for accurate cell segmentation in whole slide images using yolov11 stardist and sam2
topic WSI
cell segmentation
SAM2
deep learning
url https://www.mdpi.com/2306-5354/12/6/674
work_keys_str_mv AT juliusbamwenda ahybriddeeplearningframeworkforaccuratecellsegmentationinwholeslideimagesusingyolov11stardistandsam2
AT mehmetsiracozerdem ahybriddeeplearningframeworkforaccuratecellsegmentationinwholeslideimagesusingyolov11stardistandsam2
AT orhanayyıldız ahybriddeeplearningframeworkforaccuratecellsegmentationinwholeslideimagesusingyolov11stardistandsam2
AT veysıakpolat ahybriddeeplearningframeworkforaccuratecellsegmentationinwholeslideimagesusingyolov11stardistandsam2
AT juliusbamwenda hybriddeeplearningframeworkforaccuratecellsegmentationinwholeslideimagesusingyolov11stardistandsam2
AT mehmetsiracozerdem hybriddeeplearningframeworkforaccuratecellsegmentationinwholeslideimagesusingyolov11stardistandsam2
AT orhanayyıldız hybriddeeplearningframeworkforaccuratecellsegmentationinwholeslideimagesusingyolov11stardistandsam2
AT veysıakpolat hybriddeeplearningframeworkforaccuratecellsegmentationinwholeslideimagesusingyolov11stardistandsam2