EUVIMEDEUVIMED entdecken
Multizentrische Studie

Deep Learning Models for CT Segmentation of Invasive Pulmonary Aspergillosis, Mucormycosis, Bacterial Pneumonia and Tuberculosis: A Multicentre Study

Yun Li iD, Feifei Huang iD, Deyan Chen iD, Youwen Zhang, Xia Zhang, Lina Liang, Junnan Pan, Lunfang Tan, Shuyi Liu, Junfeng Lin, Zhengtu Li, Guodong Hu und 4 weitere

Mycoses · 2025

DOI: 10.1111/myc.70084 ↗

Kurz gefasst

Multizentrische Studie in der Zeitschrift Mycoses, 2025. Verfasst von Yun Li und 15 weiteren Autor:innen. Beteiligte Einrichtung: National Clinical Research Center for Respiratory Disease, State Key Laboratory of Respiratory Disease, Guangzhou Institute of Respiratory Health, th…. Hauptthemen laut MEDLINE: Deep Learning, Tomography, X-Ray Computed, Invasive Pulmonary Aspergillosis, Mucormycosis. Die Arbeit wurde bisher 1-mal zitiert (Crossref) und nennt 30 Literaturangaben. Gefördert von: Noncommunicable Chronic Diseases-National Science and Technology Major Project, The R&D Program of Guangzhou National Laboratory, The Science and Technology Program of Guangzhou, China.

AbstractSchlagwörterBibliografische AngabenZitierenVerwandte Studien

Abstract (Originalsprache)

Background

The differential diagnosis of invasive pulmonary aspergillosis (IPA), pulmonary mucormycosis (PM), bacterial pneumonia (BP) and pulmonary tuberculosis (PTB) are challenging due to overlapping clinical and imaging features. Manual CT lesion segmentation is time‐consuming, deep‐learning (DL)‐based segmentation models offer a promising solution, yet disease‐specific models for these infections remain underexplored.

Objectives

We aimed to develop and validate dedicated CT segmentation models for IPA, PM, BP and PTB to enhance diagnostic accuracy. Methods:Retrospective multi‐centre data (115 IPA, 53 PM, 130 BP, 125 PTB) were used for training/internal validation, with 21 IPA, 8PM, 30 BP and 31 PTB cases for external validation. Expert‐annotated lesions served as ground truth. An improved 3D U‐Net architecture was employed for segmentation, with preprocessing steps including normalisations, cropping and data augmentation. Performance was evaluated using Dice coefficients. Results:Internal validation achieved Dice scores of 78.83% (IPA), 93.38% (PM), 80.12% (BP) and 90.47% (PTB). External validation showed slightly reduced but robust performance: 75.09% (IPA), 77.53% (PM), 67.40% (BP) and 80.07% (PTB). The PM model demonstrated exceptional generalisability, scoring 83.41% on IPA data. Cross‐validation revealed mutual applicability, with IPA/PTB models achieving > 75% Dice for each other's lesions. BP segmentation showed lower but clinically acceptable performance ( >72%), likely due to complex radiological patterns.

Conclusions

Disease‐specific DL segmentation models exhibited high accuracy, particularly for PM and PTB. While IPA and BP models require refinement, all demonstrated cross‐disease utility, suggesting immediate clinical value for preliminary lesion annotation. Future efforts should enhance datasets and optimise models for intricate cases.

Schlagwörter

MeSH-Schlagwörter (MEDLINE)

Deep LearningTomography, X-Ray ComputedInvasive Pulmonary AspergillosisMucormycosisPneumonia, BacterialTuberculosis, PulmonaryHumansRetrospective StudiesDiagnosis, DifferentialMaleMiddle AgedFemaleAdultAgedLung

Stichwörter der Autor:innen

computed tomography segmentationdeep learningdice coefficientinvasive pulmonary aspergillosispulmonary mucormycosis

Autor:innen und Institutionen

  • National Clinical Research Center for Respiratory Disease, State Key Laboratory of Respiratory Disease, Guangzhou Institute of Respiratory Health, the First Affiliated Hospital of Guangzhou Medical University, Guangzhou…
  • Neusoft Research of Intelligent Healthcare Technology, Co. Ltd, Shenyang, China.
  • Neusoft Corporation, Shenyang, China.
  • Department of Neurology, Gaozhou People's Hospital, Gaozhou, China.
  • Department of Respiratory and Critical Care Medicine, The Tenth Affiliated Hospital of Southern Medical University, Guangzhou, China.
  • Department of Radiology, The Second Affiliated Hospital of Guangzhou Medical University, Guangzhou, China.

ORCID-Profile: Yun Li, Feifei Huang, Deyan Chen, Jinping Zheng

Förderung

  • Noncommunicable Chronic Diseases-National Science and Technology Major Project
  • The R&D Program of Guangzhou National Laboratory
  • The Science and Technology Program of Guangzhou, China
  • Zhongnanshan Medical Foundation of Guangdong Province
  • Science and Technology Program of Maoming, Guangdong, China
  • National Key Research and Development Program of China

Zitieren

APA

Li, Y., Huang, F., Chen, D., Zhang, Y., Zhang, X., Liang, L., Pan, J., Tan, L., Liu, S., Lin, J., Li, Z., Hu, G., Chen, H., Peng, C., Ye, F., & Zheng, J. (2025). Deep Learning Models for CT Segmentation of Invasive Pulmonary Aspergillosis, Mucormycosis, Bacterial Pneumonia and Tuberculosis: A Multicentre Study. Mycoses. https://doi.org/10.1111/myc.70084

Vancouver

Li Y, Huang F, Chen D, Zhang Y, Zhang X, Liang L, et al. Deep Learning Models for CT Segmentation of Invasive Pulmonary Aspergillosis, Mucormycosis, Bacterial Pneumonia and Tuberculosis: A Multicentre Study. Mycoses. 2025. doi:10.1111/myc.70084

BibTeX
@article{Li2025,
  author = {Li, Yun and Huang, Feifei and Chen, Deyan and Zhang, Youwen and Zhang, Xia and Liang, Lina and Pan, Junnan and Tan, Lunfang and Liu, Shuyi and Lin, Junfeng and Li, Zhengtu and Hu, Guodong and Chen, Huai and Peng, Chengbao and Ye, Feng and Zheng, Jinping},
  title = {Deep Learning Models for CT Segmentation of Invasive Pulmonary Aspergillosis, Mucormycosis, Bacterial Pneumonia and Tuberculosis: A Multicentre Study},
  journal = {Mycoses},
  year = {2025},
  doi = {10.1111/myc.70084}
}
Kostenlos starten →