Pulmonary Embolism Detection Using CNN-LSTM Hybrid Model

Pulmonary Embolism (PE) is a lifethreatening condition caused by blood clots obstructing pulmonary arteries, often leading to severe complications or death if not detected early. While CT Pulmonary Angiography (CTPA) remains the gold standard for PE diagnosis, it is time consuming and relies heavily on radiologist expertise, causing delays in critical cases. Deep learning techniques, particularly Convolutional Neural Networks (CNNs) and Long Short Term Memory (LSTMs), offer a promising solution for automating PE detection in CTPA images. By utilizing a publicly available dataset of annotated CT scans, CNNs are trained to identify embolic regions accurately. Pre-processing techniques, including image normalization and segmentation, enhance input quality, ensuring improved model performance. The system is evaluated using key metrics such as accuracy, sensitivity, specificity, and Area Under the Curve (AUC), demonstrating high detection reliability while significantly reducing diagnostic time. This approach combines speed and precision, providing radiologists with a powerful AI-driven tool to support rapid decision-making and improve patient outcomes in clinical settings. Future improvements will focus on expanding datasets, enhancing model accuracy, and integrating the solution into real-world workflows for efficient pulmonary embolism detection.