PMFF-Net: A deep learning-based image classification model for UIP, NSIP, and OP

Ming-Wei Xu; Zheng-Hua Zhang; Xiao Wang; Chun-Tao Li; Hui-Yun Yang; Ze-Hua Liao; Jian-Qing Zhang

doi:10.1016/j.compbiomed.2025.110618

PMFF-Net: A deep learning-based image classification model for UIP, NSIP, and OP

Comput Biol Med. 2025 Jun 19:195:110618. doi: 10.1016/j.compbiomed.2025.110618. Online ahead of print.

Authors

Ming-Wei Xu¹, Zheng-Hua Zhang², Xiao Wang³, Chun-Tao Li¹, Hui-Yun Yang¹, Ze-Hua Liao¹, Jian-Qing Zhang⁴

Affiliations

¹ Department of Respiratory Critical Care Medicine, The First Affiliated Hospital of Kunming Medical University, Kunming, 650032, Yunnan, People's Republic of China.
² Medical Imaging Department, The First Affiliated Hospital of Kunming Medical University, Kunming, Yunnan, 650032, People's Republic of China.
³ School of Information, Yunnan University, Kunming, Yunnan, 650032, People's Republic of China.
⁴ Department of Respiratory Critical Care Medicine, The First Affiliated Hospital of Kunming Medical University, Kunming, 650032, Yunnan, People's Republic of China. Electronic address: ydyyzjq@163.com.

PMID: 40543280
DOI: 10.1016/j.compbiomed.2025.110618

Abstract

Background: High-resolution computed tomography (HRCT) is helpful for diagnosing interstitial lung diseases (ILD), but it largely depends on the experience of physicians. Herein, our study aims to develop a deep-learning-based classification model to differentiate the three common types of ILD, so as to provide a reference to help physicians make the diagnosis and improve the accuracy of ILD diagnosis.

Methods: Patients were selected from four tertiary Grade A hospitals in Kunming based on inclusion and exclusion criteria. HRCT scans of 130 patients were included. The imaging manifestations were usual interstitial pneumonia (UIP), non-specific interstitial pneumonia (NSIP), and organizing pneumonia (OP). Additionally, 50 chest HRCT cases without imaging abnormalities during the same period were selected.Construct a data set. Conduct the training, validation, and testing of the Parallel Multi-scale Feature Fusion Network (PMFF-Net) deep learning model. Utilize Python software to generate data and charts pertaining to model performance. Assess the model's accuracy, precision, recall, and F1-score, and juxtapose its diagnostic efficacy against that of physicians across various hospital levels, with differing levels of seniority, and from various departments.

Results: The PMFF -Net deep learning model is capable of classifying imaging types such as UIP, NSIP, and OP, as well as normal imaging. In a mere 105 s, it makes the diagnosis for 18 HRCT images with a diagnostic accuracy of 92.84 %, precision of 91.88 %, recall of 91.95 %, and an F1 score of 0.9171. The diagnostic accuracy of senior radiologists (83.33 %) and pulmonologists (77.77 %) from tertiary hospitals is higher than that of internists from secondary hospitals (33.33 %). Meanwhile, the diagnostic accuracy of middle-aged radiologists (61.11 %) and pulmonologists (66.66 %) are higher than junior radiologists (38.88 %) and pulmonologists (44.44 %) in tertiary hospitals, whereas junior and middle-aged internists at secondary hospitals were unable to complete the tests.

Conclusion: This study found that the PMFF-Net model can effectively classify UIP, NSIP, OP imaging types, and normal imaging, which can help doctors of different hospital levels and departments make clinical decisions quickly and effectively.

Keywords: Deep learning; Image classification model; Interstitial lung disease; Interstitial pneumonia; Non-specific interstitial pneumonia; Organizing pneumonia; Usual.