Vessel-aligned Multi-planar Image Representation for Automated Pulmonary Embolism Detection with Convolutional Neural Networks

Description

Pulmonary embolism (PE) is a common cardiovascular emergency. Quick and accurate diagnosis of PE is critical, so that scientifically proven and efficacious life-saving treatment can be administered appropriately. Suspected PE is typically diagnosed with CT pulmonary angiography (CTPA), but even with recent increases in diagnostic accuracy, this technique still has several issues with interpretation of intricate branching structures, artifacts that may obscure or mimic embolisms, suboptimal contrast, and inhomogeneities. Computer-aided diagnosis (CAD) could play a major role in diagnosing PE, however, to achieve a clinically acceptable sensitivity, existing CAD systems generate a high number of false positives, imposing extra burdens on radiologists.

Researchers at Arizona State University have developed novel approaches for automated computer-aided detection of emboli in CTPA. One technique automatically registers the vessel orientation in display, providing compelling demonstration of arterial filling defects, if present, and allowing the radiologist to thoroughly inspect the vessel lumen from multiple perspectives and report any filling defects with high confidence. Another uses neural networks and vessel-aligned multi-planar representations to eliminate false positives.

The flexibility of these systems, coupled with their precise detection of both acute and chronic PE, significantly reduces radiologist workload and improves the efficiency and accuracy of PE diagnosis in CTPA.

Applications

•       Accurate & Automated diagnosis of PE in CTPA images

Benefits and Applications

•       Detects both acute and chronic pulmonary emboli

•       Allows visualization of vascular intensity levels and local vascular structure and occlusion

•       Navigates the vessel based on its local structure

•       Enables thorough inspection of the vessel lumen from multiple perspectives through automatic registration

•       Incrementally reports any detection to facilitate real-time support

•       Efficient and compact - concisely summarizing the 3D contextual information around an embolus

•       Consistent - automatically aligning the embolus according to its containing vessel orientation

•       Expandable—naturally supporting data augmentation for training

For more information about the inventor(s) and their research, please see
Dr. Liang’s laboratory webpage 

Case ID:
M15-186L
Published:
05-16-2016
Last Updated:
05-10-2018

Patent Information

App Type:
Provisional
Serial No.:
62/187,720
Patent No.:
File Date:
07-01-2015
Issue Date:
11-14-2018
Expire Date:
07-01-2016

For More Information, Contact