Pi-PE: A Pipeline for Pulmonary Embolism Detection using Sparsely Annotated 3D CT Images

TL;DR

This paper introduces Pi-PE, a two-stage AI pipeline for detecting pulmonary embolisms in 3D CT images that is accurate, efficient, and works with sparse annotations, showing high performance across diverse datasets.

Contribution

The paper presents a novel two-stage detection pipeline that achieves state-of-the-art results using sparse annotations and fewer model parameters, enhancing generalizability and efficiency.

Findings

Achieved AUC of 0.94 on validation and 0.85 on test sets.

Effective detection across diverse PE types and multiple hospitals.

Guidelines for designing generalizable PE detection pipelines.

Abstract

Pulmonary embolisms (PE) are known to be one of the leading causes for cardiac-related mortality. Due to inherent variabilities in how PE manifests and the cumbersome nature of manual diagnosis, there is growing interest in leveraging AI tools for detecting PE. In this paper, we build a two-stage detection pipeline that is accurate, computationally efficient, robust to variations in PE types and kernels used for CT reconstruction, and most importantly, does not require dense annotations. Given the challenges in acquiring expert annotations in large-scale datasets, our approach produces state-of-the-art results with very sparse emboli contours (at 10mm slice spacing), while using models with significantly lower number of parameters. We achieve AUC scores of 0.94 on the validation set and 0.85 on the test set of highly severe PEs. Using a large, real-world dataset characterized by complex PE types and patients from multiple hospitals, we present an elaborate empirical study and provide guidelines for designing highly generalizable pipelines.