System and method for automatic pulmonary embolism detection
A system and method for detecting pulmonary embolisms in a subject's vasculature are provided. In some aspects, the method includes acquiring a set of images representing a vasculature of the subject, and analyzing the set of images to identify pulmonary embolism candidates associated with the vasculature. The method also includes generating, for identified pulmonary embolism candidates, image patches based on a vessel-aligned image representation, and applying a set of convolutional neural networks to the generated image patches to identify pulmonary embolisms. The method further includes generating a report indicating identified pulmonary embolisms.
1. A system for detecting pulmonary embolisms in a subject's vasculature, the system comprising:
an input configured to receive images acquired from a subject;
a processor configured to process the images with steps comprising:
i. receiving, using the input, a set of images representing a vasculature of the subject;
ii. analyzing the set of images to identify pulmonary embolism candidates associated with the vasculature;
iii. generating, for identified pulmonary embolism candidates, image patches based on a vessel-aligned image representation;
iii. applying a set of convolutional neural networks to the generated image patches to identify pulmonary embolisms;
vi. generating a report indicating identified pulmonary embolisms; and
an output for displaying the report.
2. The system of claim 1 , wherein the set of images comprises computed tomography pulmonary angiography (“CTPA”) images.
3. The system of claim 1 , wherein the set of images comprises three-dimensional (“3D”) images.
4. The system of claim 1 , wherein the processor is further configured to execute a tobogganing algorithm to identify the pulmonary embolism candidates.
5. The system of claim 1 , wherein the vessel-aligned image representation comprises a longitudinal view and a cross-sectional view of vessels associated with identified pulmonary embolism candidates.
6. The system of claim 1 , wherein the processor is further configured to determine an orientation of vessels associated with identified pulmonary embolism candidates to determine the vessel-aligned image representation.
7. The system of claim 6 , wherein the processor is further configured to determine the orientation by performing vessel orientation analysis based on a principle component analysis.
8. The system of claim 6 , wherein the processor is further configured to determine the orientation by using a structure tensor or a Hessian matrix.
9. A method for detecting pulmonary embolisms in a subject's vasculature, the system comprising:
acquiring a set of images representing a vasculature of the subject;
analyzing the set of images to identify pulmonary embolism candidates associated with the vasculature;
generating, for identified pulmonary embolism candidates, image patches based on a vessel-aligned image representation;
applying a set of convolutional neural networks to the generated image patches to identify pulmonary embolisms; and
generating a report indicating identified pulmonary embolisms.
10. The method of claim 9 , wherein the set of images comprises computed tomography pulmonary angiography (“CTPA”) images.
11. The method of claim 9 , wherein the set of images comprises three-dimensional (“3D”) images.
12. The method of claim 9 , wherein the method further comprises executing a tobogganing algorithm to identify the pulmonary embolism candidates.
13. The method of claim 9 , wherein the vessel-aligned image representation comprises a longitudinal view and a cross-sectional view of vessels associated with identified pulmonary embolism candidates.
14. The method of claim 9 , wherein the method further comprises determining an orientation of vessels associated with identified pulmonary embolism candidates to determine the vessel-aligned image representation.
15. The method of claim 14 , wherein the method further comprises determining the orientation by performing a vessel orientation analysis based on a principle component analysis.
16. The method of claim 14 , wherein the method further comprises determining the orientation by performing a vessel orientation analysis using a structure tensor or a Hessian matrix.