Projects per year
Abstract / Description of output
Background
Computed Tomography (CT) is commonly used to image patients with ischemic stroke, but radiologist interpretation may be delayed. Machine learning techniques can provide rapid automated CT assessment, but are usually developed from annotated images which necessarily limits the size and representation of development datasets. We aimed to develop a deep learning (DL) method using CT brain scans that were labelled but not annotated for the presence of ischemic lesions.
Methods
We designed a convolutional neural network-based DL algorithm to detect ischemic lesions on CT. Our alogirithm was trained using routinely-aquired CT brain scans collected for a large multicentre international trial. These scans had previously been labelled by experts for acute and chronic appearances. We explored the impact of ischemic lesion features, background brain appearances, and timing of CT (baseline or 24-48 hour follow-up) on DL performance.
Results
From 5772 CT scans of 2347 patients (median age 82), 54% had visible ischemic lesions according to experts. Our DL method achieved 72% accuracy for detecting ischemic lesions. Detection was better for larger (80% accuracy) or multiple (87% accuracy for two, 100% for three or more) lesions, and with follow-up scans (76% accuracy versus 67% at baseline). Chronic brain conditions reduced accuracy, particularly non-stroke lesions and old stroke lesions (32% and 31% error rates respectively).
Conclusion
DL methods can be designed for ischemic lesion detection on CT using the vast quantities of routinely-collected brain scans without the need for lesion annotation. Ultimately, this should lead to more robust and widely applicable methods.
Computed Tomography (CT) is commonly used to image patients with ischemic stroke, but radiologist interpretation may be delayed. Machine learning techniques can provide rapid automated CT assessment, but are usually developed from annotated images which necessarily limits the size and representation of development datasets. We aimed to develop a deep learning (DL) method using CT brain scans that were labelled but not annotated for the presence of ischemic lesions.
Methods
We designed a convolutional neural network-based DL algorithm to detect ischemic lesions on CT. Our alogirithm was trained using routinely-aquired CT brain scans collected for a large multicentre international trial. These scans had previously been labelled by experts for acute and chronic appearances. We explored the impact of ischemic lesion features, background brain appearances, and timing of CT (baseline or 24-48 hour follow-up) on DL performance.
Results
From 5772 CT scans of 2347 patients (median age 82), 54% had visible ischemic lesions according to experts. Our DL method achieved 72% accuracy for detecting ischemic lesions. Detection was better for larger (80% accuracy) or multiple (87% accuracy for two, 100% for three or more) lesions, and with follow-up scans (76% accuracy versus 67% at baseline). Chronic brain conditions reduced accuracy, particularly non-stroke lesions and old stroke lesions (32% and 31% error rates respectively).
Conclusion
DL methods can be designed for ischemic lesion detection on CT using the vast quantities of routinely-collected brain scans without the need for lesion annotation. Ultimately, this should lead to more robust and widely applicable methods.
Original language | English |
---|---|
Pages (from-to) | 1-9 |
Number of pages | 9 |
Journal | Stroke and Vascular Neurology |
DOIs | |
Publication status | Published - 20 Nov 2024 |
Fingerprint
Dive into the research topics of 'Development of a deep learning method to identify acute ischemic stroke lesions on brain CT'. Together they form a unique fingerprint.Projects
- 1 Finished
-
IST3 (Supplement to R39189) - Third International Stroke Trial (Formerlly MRC G0400069)
Sandercock, P., Dennis, M. & Wardlaw, J.
1/04/10 → 30/09/13
Project: Research