Valve regurgitant detection for echocardiography
A regurgitant orifice of a valve is detected. The valve is detected from ultrasound data. An anatomical model of the valve is fit to the ultrasound data. This anatomical model may be used in various ways to assist in valvular assessment. The model may define anatomical locations about which data is sampled for quantification. The model may assist in detection of the regurgitant orifice using both B-mode and color Doppler flow data with visualization without the jet. Segmentation of a regurgitant jet for the orifice may be constrained by the model. Dynamic information may be determined based on the modeling of the valve over time.
1. A method for detecting a regurgitant point in echocardiography, the method comprising:
detecting, by a processor, a valve with a first machine-learnt classifier using input first features from both B-mode and flow-mode data for a patient;
fitting, by the processor, a model of the valve to the detected valve of the patient;
detecting, by the processor, the regurgitant point with a second machine-learnt classifier using second features from both the B-mode and the flow-mode data, the second machine-learnt classifier being applied only to locations along a free edge of the model as fit to the valve;
segmenting, by the processor, regurgitant jet image data of a regurgitant jet using the regurgitant point as a seed point; and
outputting an image including the B-mode data, the flow-mode data, the model as fit to the valve, and the regurgitant jet image data as segmented.
2. The method of claim 1 wherein detecting the valve comprises detecting a location, orientation, and scale of the valve represented by the B-mode data.
3. The method of claim 1 wherein fitting the model comprises fitting an annulus, leaflets, free edge, or combinations thereof.
4. The method of claim 1 wherein fitting the model comprises transforming a statistical shape model as a function of the detected valve.
5. The method of claim 1 wherein detecting with the first and second machine-learnt classifiers comprises calculating Haar, steerable, or Haar and steerable feature values for the B-mode data and for the flow-mode data.
6. The method of claim 1 further comprising calculating the quantity from sample planes positioned relative to the detected valve.
7. The method of claim 1 wherein segmenting comprises applying a random walker for the flow-mode data while using the regurgitant point to spatially limit the segmentation of the regurgitant jet.
8. The method of claim 1 wherein outputting further comprises outputting a quantity as representing the regurgitant jet.
9. The method of claim 1 further comprising repeating the detecting the valve, fitting, detecting the regurgitant point, and segmenting for different times in a cardiac cycle.
10. The method of claim 1 further comprising:
scanning, with a transducer adjacent the patient, a cardiac region of the patient with ultrasound;
detecting, with a B-mode detector and in response to the scanning, the B-mode data representing tissue in the cardiac region;
estimating, with a flow estimator and in response to the scanning, the flow-mode data representing fluid in the cardiac region, the flow-mode data comprising energy, velocity, or energy and velocity.
11. The method of claim 1 wherein detecting the valve comprises constraining the model as fit to the valve or part of the model as fit to the valve to locations without flow-mode data.
12. In a non-transitory computer readable storage medium having stored therein data representing instructions executable by a programmed processor for detecting a regurgitant orifice, the storage medium comprising instructions for:
detecting, at different times, anatomy of a heart valve from first features derived from both B-mode and color flow Doppler data;
sampling, at the different times, transvalvular flow over time from the color flow Doppler data on sampling planes positioned relative to the detected heart valves for the different times;
calculating a quantity of transvalvular flow from the sampling;
fitting a model of the heart valve to the detected anatomy of the heart valve;
detecting the regurgitant orifice from second features from both the color flow Doppler data and from the B-mode data, by applying a machine-learnt classifier only to locations along a free edge of the model as fit to the heart valve;
segmenting a regurgitant jet for the heart valve using the regurgitant orifice as a seed point; computing a clinical biomarker for the regurgitant jet; and
providing a visualization of the anatomy of the heart valve including the clinical biomarker and the quantity of transvalvular flow.
13. The non-transitory computer readable storage medium of claim 12 wherein detecting the anatomy comprises detecting aortic, mitral, pulmonary, or tricuspid anatomy.
14. The non-transitory computer readable storage medium of claim 12 further comprising computing dynamic flow quantity for flow as a function of time from the sampling.
15. The non-transitory computer readable storage medium of claim 12 further comprising displaying the model as fit to the heart valve and the regurgitant orifice without the regurgitant jet.
16. A system for detecting a regurgitant region, the system comprising:
an ultrasound scanner configured to scan a heart volume of a patient, the scan providing B-mode and Doppler flow data;
a processor configured to fit a model of a heart valve over time to the B-mode data using the B-mode data and the Doppler flow data, and use the model as fit to the heart valve to locate the regurgitant region over time by applying a machine-learnt classifier only to locations along a free edge of the model as fit to the heart valve; and a display configured to generate a visualization of the model as fit to the heart valve over time and highlight the regurgitant region without displaying flow data for a regurgitant jet.