کنفرانس مهندسی برق ایران

صفحه اصلی / سی و سومین کنفرانس بین المللی مهندسی برق

A Novel Approach to Pulmonary Embolism Segmentation: Increasing an Attention-based U-Net

نویسندگان :

Hanie Arabian¹ Alireza Karimian² Hosein Arabi³ Marjan Mansourian⁴

1- دانشگاه اصفهان 2- دانشگاه اصفهان 3- University of Geneva 4- دانشگاه علوم پزشکی اصفهان

کلمات کلیدی :

deep learning،image segmentation،pulmonary embolism،squeeze-and-attention block،u-net architecture

چکیده :

Pulmonary embolism (PE) is a life-threatening condition, often leading to late diagnoses. Diagnostic tools like Computed Tomography Pulmonary Angiography (CTPA) rely on radiologist skills, resulting in variable sensitivity and specificity. This study aims to leverage deep learning, specifically a convolutional neural network with U-net architecture enhanced by Squeeze-and-Attention and Long Short-Term Memory (LSTM) blocks, to improve the segmentation of emboli in CTPA images. Utilizing two datasets, CAD-PE (91 cases, 89 with PE) and FUMPE (35 cases, 33 with PE), the research assesses how increasing the number of network layers (57, 67, and 103) affects segmentation performance. The results demonstrated that the slice-wise sensitivity improved from 76.73±21.94 with a 57-layer architecture to 80.36±21.42 with a 67-layer architecture, indicating better pulmonary embolism detection (with a significant difference due to paired T-test P-value of less than 0.05). In addition, the patient-wise AUC slightly increases from 81.68±10.94 (57 layers) to 85.09±10.69 (67 layers) with a Kruskal-Wallis P-value of 0.0189, which indicates a significant difference between the networks’ performance. However, no significant improvement was observed with the 103-layer model, highlighting the potential for overfitting. Results from this study demonstrate the potential of deep learning algorithms in enhancing the accurate diagnosis of pulmonary embolism. Finally, the neural network's performance in segmenting pulmonary embolisms from CT images demonstrates promising directions with particular specificity and overall AUC strengths.

لیست مقالات

لیست مقالات بایگانی شده

Age of Information Optimization for Multi-hop VLC/RF IoT Sensor Networks

Hossein Khodi - Paeiz Azmi - Nader Mokari - Mohammadreza Javan - Hamid Saeedi - Murat Uysal

Towards Non-Invasive Deep Brain Stimulation Using Temporal Interference Method

Mehdi Gholami - Farshid Ghobadzadeh - Fatemeh Yazdanshenas - Amir Yazdani - Mohammad Neshat

مدیریت انرژی شارژر خودروهای الکتریکی، به منظور افزایش ضریب نفوذ خودروهای الکتریکی و بهبود پروفیل ولتاژ شبکه های توزیع الکتریکی هوشمند با استفاده از شارژ خودروها در محل کار

مهدی افشار - سعید اسماعیلی جعفرآبادی

Revealing Shadows: Low-Light Image Enhancement Using Self-Calibrated Illumination

Farzaneh Koohestani - Nader Karimi - Shadrokh Samavi

Microgrid Damping Improvement Using High-Pass Filter-Based Virtual Synchronous Generator

Shayan Zaimi - Ashkan Moradi Naserkhani - Sharara Rehimi - Amin Karimi - Rahmatollah Mirzaei - Hassan Bevrani

Brain Effective Connectivity Comparision in Different States of Familiarity and Desiring Brands Confrontation: a Neuromarketing Study

Mahdi Taghaddossi - Mohammad Hasan Moradi

Improved Attention U-Net combined with Conditional Random Field for Ischemic Lesion Segmentation from Magnetic Resonance Images

Ali Rezaei - Asieh Khosravanian - Habibollah Danyali - Kamran Kazemi - Ardalan Aarabi

On spatiotemporal-aware deep neural networks for real-time video fire detection: empowering image-based models with temporal and spatial features of video

Mahdi Shamisavi - Sahar Eslami - Amir Jahanshahi - Morteza Rajabzadeh

Deep Convolutional Neural Network for ADHD Classification using resting-state fMRI

MohammadHadi Firouzi - Maliheh Ahmadi - Kamran Kazemi - Mohammad Sadegh Helfroush - Ardalan Aarabi

Three-Winding Coupled-Inductor-based Boost Converter with Voltage Multiplier Cell and Active Clamp Circuit for Low-Power Photovoltaic Application

Danesh Amani - Ali Valizadeh - Reza Beiranvand - Ali Yazdian Varjani

بیشتر