Optimization of Multiresolution Sinusoidal Neural Network for CT Imaging of Pulmonary Embolism for Heart Angiography Identification
Year: 2026
Type: conference-paper
Abstract: The detection of pulmonary embolism (PE) and coronary disease abnormalities is a time-sensitive procedure in cardiopulmonary emergencies and needs accurate analysis of CT scans. In this work, a novel Multiresolution Sinusoidal Neural Network (MSNN) is introduced which can simultaneously detect pulmonary embolism and cardiac angiography abnormalities using a single framework based on CT scan analysis. The network adopts the use of a multiresolution hierarchical structure alongside sinusoidal activation functions in order to identify subtle abnormalities in vessels and emboli that could not be detected by standard CNN architectures. In order to further enhance the diagnosis, adaptive learning rates, regularization and attention-based feature orientation mechanisms are applied. Results from experimentation reveal the capability of the proposed MSNN in achieving 96.87% accuracy and 98.75% AUC score compared to state-of-the-art models like EfficientNet-B7, DFU Care, ResNet-50, and Vision Transformer in a highly efficient way of processing. The framework helps radiologists in fast interpretation of images and emergency triage planning.
Language: English
Topic: COVID-19 diagnosis using AI
Subfield: Radiology, Nuclear Medicine and Imaging
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W7225387178