HG-YOLO: Improving Tumor Detection with PP-HGNet and Global Attention Mechanism

Kien Trang; Bao Quoc Vuong; An Hoang Nguyen; Fung Fung Ting; Chee Ming Ting

doi:10.1007/978-3-031-97000-9_21

HG-YOLO: Improving Tumor Detection with PP-HGNet and Global Attention Mechanism

Kien Trang, Bao Quoc Vuong, An Hoang Nguyen, Fung Fung Ting, Chee Ming Ting

Malaysia Sch of Information Technology

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

Abstract

Cancer continues to be a major global health issue, which is defined by uncontrolled cell proliferation and the potential invasion to other areas of the body. The early and precise detection of tumors through medical imaging is essential in improving cancer prognosis and treatment outcomes. This study presents HG-YOLO, a novel YOLO-based architecture enhanced by the integration of the PP-HGNet backbone and a Global Attention Mechanism (GAM). HG-YOLO capitalizes on the robust feature extraction capabilities of PP-HGNet and the attention-enhancing properties of GAM to improve the potential features. This combination aims to improve the sensitivity and precision of tumor localization, especially in complex cases where tumors are small or poorly delineated. Overall, the model is assessed using the Brain Tumor Detection 2020 (Br35H) dataset with the Magnetic Resonance Imaging (MRI) images. Comparative studies show that our HG-YOLO outperforms the prior versions in terms of Precision, Recall, mAP50 and mAP50-95 - giving 0.934, 0.915, 0.953, and 0.728, respectively.

Original language	English
Title of host publication	Computational Science and Its Applications – ICCSA 2025 - 25th International Conference Istanbul, Turkey, June 30 – July 3, 2025 Proceedings, Part I
Editors	Osvaldo Gervasi, Beniamino Murgante, Chiara Garau, Yeliz Karaca, David Taniar, Ana Maria A. C. Rocha, Bernady O. Apduhan
Place of Publication	Cham Switzerland
Publisher	Springer
Pages	332-344
Number of pages	13
ISBN (Electronic)	9783031970009
ISBN (Print)	9783031969997
DOIs	https://doi.org/10.1007/978-3-031-97000-9_21
Publication status	Published - 2025
Event	International Conference on Computational Science and Applications 2025 - Istanbul, Türkiye Duration: 30 Jun 2025 → 3 Jul 2025 Conference number: 25th https://link.springer.com/book/10.1007/978-3-031-97000-9 (Proceedings) https://iccsa.org/ (Website)

Publication series

Name	Lecture Notes in Computer Science
Publisher	Springer
Volume	15648
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	International Conference on Computational Science and Applications 2025
Abbreviated title	ICCSA 2025
Country/Territory	Türkiye
City	Istanbul
Period	30/06/25 → 3/07/25
Internet address	https://link.springer.com/book/10.1007/978-3-031-97000-9 (Proceedings) https://iccsa.org/ (Website)

Keywords

brain tumor
deep learning
MRI
PP-HGNet
YOLO

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1007/978-3-031-97000-9_21

Cite this

Trang, K., Vuong, B. Q., Nguyen, A. H., Ting, F. F., & Ting, C. M. (2025). HG-YOLO: Improving Tumor Detection with PP-HGNet and Global Attention Mechanism. In O. Gervasi, B. Murgante, C. Garau, Y. Karaca, D. Taniar, A. M. A. C. Rocha, & B. O. Apduhan (Eds.), Computational Science and Its Applications – ICCSA 2025 - 25th International Conference Istanbul, Turkey, June 30 – July 3, 2025 Proceedings, Part I (pp. 332-344). (Lecture Notes in Computer Science; Vol. 15648). Springer. https://doi.org/10.1007/978-3-031-97000-9_21

Trang, Kien ; Vuong, Bao Quoc ; Nguyen, An Hoang et al. / HG-YOLO : Improving Tumor Detection with PP-HGNet and Global Attention Mechanism. Computational Science and Its Applications – ICCSA 2025 - 25th International Conference Istanbul, Turkey, June 30 – July 3, 2025 Proceedings, Part I. editor / Osvaldo Gervasi ; Beniamino Murgante ; Chiara Garau ; Yeliz Karaca ; David Taniar ; Ana Maria A. C. Rocha ; Bernady O. Apduhan. Cham Switzerland : Springer, 2025. pp. 332-344 (Lecture Notes in Computer Science).

@inproceedings{5be0a767dbed47d49e4c0b4b31e9634f,

title = "HG-YOLO: Improving Tumor Detection with PP-HGNet and Global Attention Mechanism",

abstract = "Cancer continues to be a major global health issue, which is defined by uncontrolled cell proliferation and the potential invasion to other areas of the body. The early and precise detection of tumors through medical imaging is essential in improving cancer prognosis and treatment outcomes. This study presents HG-YOLO, a novel YOLO-based architecture enhanced by the integration of the PP-HGNet backbone and a Global Attention Mechanism (GAM). HG-YOLO capitalizes on the robust feature extraction capabilities of PP-HGNet and the attention-enhancing properties of GAM to improve the potential features. This combination aims to improve the sensitivity and precision of tumor localization, especially in complex cases where tumors are small or poorly delineated. Overall, the model is assessed using the Brain Tumor Detection 2020 (Br35H) dataset with the Magnetic Resonance Imaging (MRI) images. Comparative studies show that our HG-YOLO outperforms the prior versions in terms of Precision, Recall, mAP50 and mAP50-95 - giving 0.934, 0.915, 0.953, and 0.728, respectively.",

keywords = "brain tumor, deep learning, MRI, PP-HGNet, YOLO",

author = "Kien Trang and Vuong, \{Bao Quoc\} and Nguyen, \{An Hoang\} and Ting, \{Fung Fung\} and Ting, \{Chee Ming\}",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2025.; International Conference on Computational Science and Applications 2025, ICCSA 2025 ; Conference date: 30-06-2025 Through 03-07-2025",

year = "2025",

doi = "10.1007/978-3-031-97000-9\_21",

language = "English",

isbn = "9783031969997",

series = "Lecture Notes in Computer Science",

publisher = "Springer",

pages = "332--344",

editor = "Osvaldo Gervasi and Beniamino Murgante and Chiara Garau and Yeliz Karaca and David Taniar and \{C. Rocha\}, \{Ana Maria A.\} and Apduhan, \{Bernady O.\}",

booktitle = "Computational Science and Its Applications – ICCSA 2025 - 25th International Conference Istanbul, Turkey, June 30 – July 3, 2025 Proceedings, Part I",

address = "Switzerland",

url = "https://link.springer.com/book/10.1007/978-3-031-97000-9, https://iccsa.org/",

}

Trang, K, Vuong, BQ, Nguyen, AH, Ting, FF & Ting, CM 2025, HG-YOLO: Improving Tumor Detection with PP-HGNet and Global Attention Mechanism. in O Gervasi, B Murgante, C Garau, Y Karaca, D Taniar, AMA C. Rocha & BO Apduhan (eds), Computational Science and Its Applications – ICCSA 2025 - 25th International Conference Istanbul, Turkey, June 30 – July 3, 2025 Proceedings, Part I. Lecture Notes in Computer Science, vol. 15648, Springer, Cham Switzerland, pp. 332-344, International Conference on Computational Science and Applications 2025, Istanbul, Türkiye, 30/06/25. https://doi.org/10.1007/978-3-031-97000-9_21

HG-YOLO: Improving Tumor Detection with PP-HGNet and Global Attention Mechanism. / Trang, Kien; Vuong, Bao Quoc; Nguyen, An Hoang et al.
Computational Science and Its Applications – ICCSA 2025 - 25th International Conference Istanbul, Turkey, June 30 – July 3, 2025 Proceedings, Part I. ed. / Osvaldo Gervasi; Beniamino Murgante; Chiara Garau; Yeliz Karaca; David Taniar; Ana Maria A. C. Rocha; Bernady O. Apduhan. Cham Switzerland: Springer, 2025. p. 332-344 (Lecture Notes in Computer Science; Vol. 15648).

Research output: Chapter in Book/Report/Conference proceeding › Conference Paper › Research › peer-review

TY - GEN

T1 - HG-YOLO

T2 - International Conference on Computational Science and Applications 2025

AU - Trang, Kien

AU - Vuong, Bao Quoc

AU - Nguyen, An Hoang

AU - Ting, Fung Fung

AU - Ting, Chee Ming

N1 - Conference code: 25th

PY - 2025

Y1 - 2025

N2 - Cancer continues to be a major global health issue, which is defined by uncontrolled cell proliferation and the potential invasion to other areas of the body. The early and precise detection of tumors through medical imaging is essential in improving cancer prognosis and treatment outcomes. This study presents HG-YOLO, a novel YOLO-based architecture enhanced by the integration of the PP-HGNet backbone and a Global Attention Mechanism (GAM). HG-YOLO capitalizes on the robust feature extraction capabilities of PP-HGNet and the attention-enhancing properties of GAM to improve the potential features. This combination aims to improve the sensitivity and precision of tumor localization, especially in complex cases where tumors are small or poorly delineated. Overall, the model is assessed using the Brain Tumor Detection 2020 (Br35H) dataset with the Magnetic Resonance Imaging (MRI) images. Comparative studies show that our HG-YOLO outperforms the prior versions in terms of Precision, Recall, mAP50 and mAP50-95 - giving 0.934, 0.915, 0.953, and 0.728, respectively.

AB - Cancer continues to be a major global health issue, which is defined by uncontrolled cell proliferation and the potential invasion to other areas of the body. The early and precise detection of tumors through medical imaging is essential in improving cancer prognosis and treatment outcomes. This study presents HG-YOLO, a novel YOLO-based architecture enhanced by the integration of the PP-HGNet backbone and a Global Attention Mechanism (GAM). HG-YOLO capitalizes on the robust feature extraction capabilities of PP-HGNet and the attention-enhancing properties of GAM to improve the potential features. This combination aims to improve the sensitivity and precision of tumor localization, especially in complex cases where tumors are small or poorly delineated. Overall, the model is assessed using the Brain Tumor Detection 2020 (Br35H) dataset with the Magnetic Resonance Imaging (MRI) images. Comparative studies show that our HG-YOLO outperforms the prior versions in terms of Precision, Recall, mAP50 and mAP50-95 - giving 0.934, 0.915, 0.953, and 0.728, respectively.

KW - brain tumor

KW - deep learning

KW - MRI

KW - PP-HGNet

KW - YOLO

UR - http://www.scopus.com/inward/record.url?scp=105010187941&partnerID=8YFLogxK

U2 - 10.1007/978-3-031-97000-9_21

DO - 10.1007/978-3-031-97000-9_21

M3 - Conference Paper

AN - SCOPUS:105010187941

SN - 9783031969997

T3 - Lecture Notes in Computer Science

SP - 332

EP - 344

BT - Computational Science and Its Applications – ICCSA 2025 - 25th International Conference Istanbul, Turkey, June 30 – July 3, 2025 Proceedings, Part I

A2 - Gervasi, Osvaldo

A2 - Murgante, Beniamino

A2 - Garau, Chiara

A2 - Karaca, Yeliz

A2 - Taniar, David

A2 - C. Rocha, Ana Maria A.

A2 - Apduhan, Bernady O.

PB - Springer

CY - Cham Switzerland

Y2 - 30 June 2025 through 3 July 2025

ER -

Trang K, Vuong BQ, Nguyen AH, Ting FF , Ting CM. HG-YOLO: Improving Tumor Detection with PP-HGNet and Global Attention Mechanism. In Gervasi O, Murgante B, Garau C, Karaca Y, Taniar D, C. Rocha AMA, Apduhan BO, editors, Computational Science and Its Applications – ICCSA 2025 - 25th International Conference Istanbul, Turkey, June 30 – July 3, 2025 Proceedings, Part I. Cham Switzerland: Springer. 2025. p. 332-344. (Lecture Notes in Computer Science). doi: 10.1007/978-3-031-97000-9_21

HG-YOLO: Improving Tumor Detection with PP-HGNet and Global Attention Mechanism

Abstract

Publication series

Conference

Keywords

UN SDGs

Access to Document

Other files and links

Cite this