Detecting and Classifying Pulmonary Embolism Using MobileNet With Sliding Window Fusion Algorithm

There are numerous methods for identifying and categorizing pulmonary embolism. When detecting Pulmonary Embolism (PE) in scans manual, semi-automated, and automatic methods based on Convolutional Neural Networks (CNN) are typically employed. Here, we segregate pulmonary vessels using a Sliding Wind...

Full description

Saved in:
Bibliographic Details
Published in2024 International Conference on Recent Advances in Electrical, Electronics, Ubiquitous Communication, and Computational Intelligence (RAEEUCCI) pp. 1 - 5
Main Authors Sekhar, Ambika, Suresh, L. Padma
Format Conference Proceeding
LanguageEnglish
Published IEEE 17.04.2024
Subjects
Classification algorithms
Computational intelligence
Convolutional neural networks
Deep learning
Lung
Manuals
Measurement
MobileNet
Pulmonary diseases
Pulmonary Embolism
Sliding Window
Online AccessGet full text

Cover

More Information
Summary:There are numerous methods for identifying and categorizing pulmonary embolism. When detecting Pulmonary Embolism (PE) in scans manual, semi-automated, and automatic methods based on Convolutional Neural Networks (CNN) are typically employed. Here, we segregate pulmonary vessels using a Sliding Window Fusion algorithm based on MobileNet to identify pulmonary ailments. When utilizing the suggested method with the FUMPE dataset, the accuracy is 99%, and when using the RSNA dataset, it is more than 90%. With the RSNA dataset, precision, recall, and F1 score were .96, .91 and .91 while with the FUMPE dataset, the results were 1.0, .95 and .96. The suggested strategy employing the RSNA and FUMPE datasets yielded an AUC score of .90 and .97 respectively.
DOI:10.1109/RAEEUCCI61380.2024.10547756