AUTOMATED IMAGE ANALYSIS FOR IDENTIFYING A MEDICAL PARAMETER
Aspects of the technology described herein relate to techniques for guiding an operator to use an ultrasound device. Thereby, operators with little or no experience operating ultrasound devices may capture medically relevant ultrasound images and/or interpret the contents of the obtained ultrasound images. For example, some of the techniques disclosed herein may be used to identify a particular anatomical view of a subject to image with an ultrasound device, guide an operator of the ultrasound device to capture an ultrasound image of the subject that contains the particular anatomical view, and/or analyze the captured ultrasound image to identify medical information about the subject.
1 . An apparatus, comprising:
at least one processor configured to:
obtain an ultrasound image of a subject; and
identify at least one medical parameter of the subject at least in part by analyzing the ultrasound image using a deep learning technique.
2 . The apparatus of claim 1 , wherein the at least one processor is configured to identify the at least one medical parameter of the subject at least in part by identifying at least one anatomical feature of the subject in the ultrasound image using the deep learning technique.
3 . The apparatus of claim 2 , wherein the at least one processor is configured to identify the at least one anatomical feature of the subject at least in part by providing the ultrasound image as an input to a multi-layer neural network.
4 . The apparatus of claim 3 , wherein the at least one processor is configured to identify the at least one anatomical feature of the subject at least in part by using the multi-layer neural network to obtain an output that is indicative of the at least one anatomical feature of the subject in the ultrasound image.
5 . The apparatus of claim 2 , wherein the at least one processor is configured to identify the at least one anatomical feature of the subject at least in part by analyzing the ultrasound image using a multi-layer neural network comprising at least one layer selected from the group consisting of: a pooling layer, a rectified linear units (ReLU) layer, a convolution layer, a dense layer, a pad layer, a concatenate layer, and an upscale layer.
6 . The apparatus of claim 2 , wherein the at least one anatomical feature comprises an anatomical feature selected from the group consisting of: a heart ventricle, a heart valve, a heart septum, a heart papillary muscle, a heart atrium, an aorta, and a lung.
7 . The apparatus of claim 1 , wherein the at least one medical parameter comprises a medical parameter selected from the group consisting of: an ejection fraction, a fractional shortening, a ventricle diameter, a ventricle volume, an end-diastolic volume, an end-systolic volume, a cardiac output, a stroke volume, an intraventricular septum thickness, a ventricle wall thickness, and a pulse rate.
8 . The apparatus of claim 1 , wherein the at least one processor is configured to overlay the at least one medical parameter onto the ultrasound image of the subject to form a composite image.
9 . The apparatus of claim 8 , further comprising a display coupled to the at least one processor and configured to display the composite image to the operator.
10 . The apparatus of claim 9 , wherein the display and the at least one processor are integrated into a computing device.
11 . A method, comprising:
using at least one computing device comprising at least one processor to perform:
obtaining an ultrasound image of a subject captured by an ultrasound device;
identifying at least one anatomical feature of the subject in the ultrasound image using an automated image processing technique; and
identifying at least one medical parameter of the subject using the identified anatomical feature in the ultrasound image.
12 . The method of claim 11 , wherein identifying the at least one anatomical feature of the subject comprises analyzing the ultrasound image using a deep learning technique.
13 . The method of claim 11 , wherein identifying the at least one anatomical feature of the subject comprises providing the ultrasound image as an input to a multi-layer neural network.
14 . The method of claim 13 , wherein identifying the at least one anatomical feature of the subject comprises using the multi-layer neural network to obtain an output that is indicative of the at least one anatomical feature of the subject in the ultrasound image.
15 . The method of claim 11 , wherein identifying the at least one anatomical feature of the subject comprises analyzing the ultrasound image using a multi-layer neural network comprising at least one layer selected from the group consisting of: a pooling layer, a rectified linear units (ReLU) layer, a convolution layer, a dense layer, a pad layer, a concatenate layer, and an upscale layer.
16 . The method of claim 11 , wherein identifying the at least one anatomical feature comprises identifying an anatomical feature selected from the group consisting of: a heart ventricle, a heart atrium, an aorta, and a lung.
17 . The method of claim 11 , wherein identifying the at least one medical parameter comprises identifying a medical parameter selected from the group consisting of: an ejection fraction, a fractional shortening, a ventricle diameter, a ventricle volume, an end-diastolic volume, an end-systolic volume, a cardiac output, a stroke volume, an intraventricular septum thickness, a ventricle wall thickness, and a pulse rate.
18 . The method of claim 11 , wherein obtaining the ultrasound image of the subject comprises obtaining a plurality of ultrasound images of the subject, and wherein identifying the at least one anatomical feature of the subject comprises identifying a ventricle in each of at least some of the plurality of ultrasound images using a multi-layer neural network.
19 . The method of claim 18 , wherein identifying the at least one medical parameter comprises:
estimating a ventricle diameter of the identified ventricles in each of the at least some of the plurality of images to obtain a plurality of ventricle diameters including a first ventricle diameter and a second ventricle diameter that is different from the first ventricle diameter;
using the first ventricle diameter to estimate an end-diastolic volume; and
using the second ventricle diameter to estimate an end-systolic volume.
20 . The method of claim 19 , wherein identifying the at least one medical parameter comprises identifying an ejection fraction of the subject using the estimated end-diastolic volume and the estimated end-systolic volume.
21 . The method of claim 11 , further comprising:
overlaying the at least one medical parameter onto the ultrasound image to form a composite image; and
presenting the composite image.
22 . The method of claim 11 , wherein obtaining the ultrasound image comprises guiding an operator of the ultrasound device to capture the ultrasound image of the subject.
23 . The method of claim 22 , wherein guiding the operator of the ultrasound device comprises providing the ultrasound image as an input to a first multi-layer neural network and wherein identifying the at least one anatomical feature of the subject comprises providing the ultrasound image as an input to a second multi-layer neural network that is different from the first multi-layer neural network.
24 . A system, comprising:
an ultrasound device configured to capture an ultrasound image of a subject; and
a computing device communicatively coupled to the ultrasound device and configured to:
obtain the ultrasound image captured by the ultrasound device;
identify at least one anatomical feature of the subject in the ultrasound image using an automated image processing technique; and
identify at least one medical parameter of the subject using the identified anatomical feature in the ultrasound image.
25 . The system of claim 24 , wherein the ultrasound device comprises a plurality of ultrasonic transducers.
26 . The system of claim 25 , wherein the plurality of ultrasonic transducers comprises an ultrasonic transducer selected from the group consisting of: a capacitive micromachined ultrasonic transducer (CMUT), a CMOS ultrasonic transducer (CUT), and a piezoelectric micromachined ultrasonic transducer (PMUT).
27 . The system of claim 24 , wherein the computing device is a mobile smartphone or a tablet.
28 . The system of claim 24 , wherein the computing device comprises a display, and wherein the computing device is configured to display an indication of the at least one medical parameter using the display.
29 . The system of claim 24 , wherein the ultrasound image contains an anatomical view selected from the group consisting of: a parasternal long axis (PLAX) anatomical view, a parasternal short-axis (PSAX) anatomical view, an apical four-chamber (A4C) anatomical view, and apical long axis (ALAX) anatomical view.
30 . At least one non-transitory computer-readable storage medium storing processor-executable instructions that, when executed by at least one processor, cause the at least one processor to:
obtain an ultrasound image of a subject captured by an ultrasound device;
identify at least one anatomical feature of the subject in the ultrasound image using an automated image processing technique; and
identify at least one medical parameter of the subject using the identified anatomical feature in the ultrasound image.