Predicting Thrombectomy Recanalization from CT Imaging Using Deep Learning Models

Haoyue Zhang,Jennifer S. Polson,Eric J. Yang,Kambiz Nael,William Speier,Corey W. Arnold

2024-04-18

Abstract:For acute ischemic stroke (AIS) patients with large vessel occlusions, clinicians must decide if the benefit of mechanical thrombectomy (MTB) outweighs the risks and potential complications following an invasive procedure. Pre-treatment computed tomography (CT) and angiography (CTA) are widely used to characterize occlusions in the brain vasculature. If a patient is deemed eligible, a modified treatment in cerebral ischemia (mTICI) score will be used to grade how well blood flow is reestablished throughout and following the MTB procedure. An estimation of the likelihood of successful recanalization can support treatment decision-making. In this study, we proposed a fully automated prediction of a patient's recanalization score using pre-treatment CT and CTA imaging. We designed a spatial cross attention network (SCANet) that utilizes vision transformers to localize to pertinent slices and brain regions. Our top model achieved an average cross-validated ROC-AUC of 77.33 $\pm$ 3.9\%. This is a promising result that supports future applications of deep learning on CT and CTA for the identification of eligible AIS patients for MTB.

Image and Video Processing,Computer Vision and Pattern Recognition

What problem does this paper attempt to address?

The paper aims to address the issue of predicting the success rate of blood flow recanalization in patients with Acute Ischemic Stroke (AIS) before undergoing Mechanical Thrombectomy (MTB). Specifically: 1. **Clinical Background**: AIS is one of the most common types of stroke, and MTB is an effective treatment for large vessel occlusion. However, when deciding whether to perform MTB, doctors need to assess whether the benefits of the surgery outweigh the potential risks and complications. 2. **Existing Challenges**: Currently, clinicians typically use preoperative Computed Tomography (CT) and Computed Tomography Angiography (CTA) to assess the condition of cerebral vessel occlusion. If the patient meets the criteria for MTB, the degree of blood flow restoration is evaluated based on the postoperative modified Treatment In Cerebral Ischemia (mTICI) score. However, predicting the success rate of MTB in advance remains challenging. 3. **Research Objective**: The paper proposes a fully automated deep learning model that uses preoperative CT and CTA imaging data to predict the patient's mTICI score, thereby supporting clinical decision-making. The model, named SCANet (Spatial Cross Attention Network), incorporates Vision Transformers technology to locate critical slices and brain regions. 4. **Main Contribution**: Experimental results show that the SCANet model achieved an average ROC-AUC of 77.33% in cross-validation, significantly outperforming existing fully automated and semi-automated machine learning/deep learning models. Moreover, it achieves high prediction accuracy without the need for manual segmentation of the thrombus area. This lays the foundation for the future development of MTB patient screening tools based on CT/CTA imaging.

Predicting Thrombectomy Recanalization from CT Imaging Using Deep Learning Models

Deep learning for prediction of post-thrombectomy outcomes based on admission CT angiography in large vessel occlusion stroke

A Deep Learning Approach to Predict Recanalization First-Pass Effect following Mechanical Thrombectomy in Patients with Acute Ischemic Stroke

Prediction of large vessel occlusion for ischaemic stroke by using the machine learning model random forests

Prediction of tissue outcome in acute ischemic stroke based on single-phase CT angiography at admission

Ensemble Learning Accurately Predicts the Potential Benefits of Thrombolytic Therapy in Acute Ischemic Stroke.

Machine Learning-Enabled Automated Determination of Acute Ischemic Core From Computed Tomography Angiography.

U-net Models Based on Computed Tomography Perfusion Predict Tissue Outcome in Patients with Different Reperfusion Patterns

Deep learning on pre-procedural computed tomography and clinical data predicts outcome following stroke thrombectomy

Abstract TP144: Deep Learning on Pre-Procedural CT and Clinical Data Predicts Outcome Following Stroke Thrombectomy

Automated identification of thrombectomy amenable vessel occlusion on computed tomography angiography using deep learning

Deep learning-based model for difficult transfemoral access prediction compared with human assessment in stroke thrombectomy

Deep learning using one-stop-shop CT scan to predict hemorrhagic transformation in stroke patients undergoing reperfusion therapy: A multicenter study

Segmentation of Infarct Lesions and Prognosis Prediction for Acute Ischemic Stroke Using Non-Contrast CT Scans

Prediction of final infarct volume from native CT perfusion and treatment parameters using deep learning

Target-based deep learning network surveillance of non-contrast computed tomography for small infarct core of acute ischemic stroke

Deep Learning-based Prediction of Percutaneous Recanalization in Chronic Total Occlusion Using Coronary CT Angiography.

Automated delineation of acute ischemic stroke lesions on non-contrast CT using 3D deep learning: A promising step towards efficient diagnosis and treatment

Abstract No. 720 Identification of Irreversibly Damaged Brain Tissue on Computed Tomography Perfusion Using Convolutional Neural Network to Assist Selection for Mechanical Thrombectomy in Ischemic Stroke Patients

Deep learning-based identification of acute ischemic core and deficit from non-contrast CT and CTA

Using Deep-Learning-Based Artificial Intelligence Technique to Automatically Evaluate the Collateral Status of Multiphase CTA in Acute Ischemic Stroke