Fetal statistic estimation in blind sweep ultrasound scans

Abstract

In some examples, fetal growth and fetal size may be detected and estimated based on data corresponding to at least one blind sweep ultrasound scan. Using a first neural network, one or more anatomies may be identified in the data corresponding to the at least one blind sweep ultrasound scan. In some examples, a bounding box may be applied to identified anatomies and various measurements and calculations may be performed on the bounding boxes. In some other examples, standard plane measurements may be performed if a standard plane is detected in the data corresponding to the at least one blind sweep. In some examples, a second neural network may determine a gestational age based on the at least one blind sweep ultrasound scan data. In some examples, at least one fetal statistic may be estimated based on the one or more anatomies and the gestational age.

Description

2024PF002871FETAL STATISTIC ESTIMATION IN BLIND SWEEP ULTRASOUND SCANSTECHNICAL FIELD

[0001] The present disclosure relates to acquiring and analyzing medical imaging data. For example, embodiments herein relate to acquiring, analyzing, and estimating various fetal statistics (e.g. growth) based on blind sweep ultrasound imaging data using neural network model(s).BACKGROUND

[0002] Various procedures for the estimation of fetal size in standard clinical workflow may be performed using the measurements of the fetus performed along many standard planes (that is, images acquired at standard planes) such as the head circumference measurement, abdominal circumference measurement, femur length and biparietal diameter of the head. The measurements along the standard planes may be compared against, for example, a Hadlock table containing different percentile ranges of a fetal size based on the standard plane measurements for a particular gestational age. However, in some maternal care settings, skilled imaging clinicians, such as sonographers, may not be readily available. Accordingly, it may be difficult to acquire images from the standard planes from which the desired fetal measurements can be obtained.

[0003] In an attempt to improve access to prenatal imaging, blind sweep ultrasound scans may be performed. These scans may be performed by a healthcare provider with little or no ultrasound imaging training. The healthcare provider makes multiple horizontal and vertical sweeps along the mother’s abdomen to obtain a set of ultrasound images. Often, the healthcare provider has no access to the images, hence the term “blind sweep.” The ultrasound images may then be provided to a radiologist or other clinician for analysis.

[0004] While blind sweep ultrasound scans may provide clinically useful data, blind sweep ultrasound scans do not always obtain data from which images from standard planes can be extracted. Accordingly, in some cases, when using blind sweeping ultrasound techniques, the estimation of a fetus size or whether a fetus is small for gestational age (SGA) or large for gestational age (LGA) may not be possible using standard techniques.SUMMARY

[0005] The invention is defined by the independent claims. The dependent claims define advantageous embodiments.

[0006] Apparatuses, systems, and methods for detecting fetal growth and estimate fetal size, are discussed herein. In some implementations, data corresponding to at least one blind sweep ultrasound scan may be acquired using a blind sweep protocol. Then, using a first neural network, one or more2024PF002872 anatomies may be identified in the data corresponding to the at least one blind sweep ultrasound scan. In some examples, a bounding box may be applied to identified anatomies and various measurements and calculations may be performed on the bounding boxes. In some other examples, measurements may be performed if an image from standard plane is detected in the data. Additionally, using a second neural network, a gestational age may be determined based on the blind sweep ultrasound scan data. Subsequently, at least one fetal statistic may be estimated based on the one or more anatomies and the gestational age.

[0007] In accordance with at least one example disclosed herein, an ultrasound imaging system is disclosed. The ultrasound imaging system comprises at least one processor configured to receive, data corresponding to at least one blind sweep ultrasound scan. The at least one processor is further configured to identify, using a first neural network, one or more anatomies in the data corresponding to the at least one blind sweep ultrasound scan. The at least one processor is further configured to determine, using a second neural network, a gestational age based on the data corresponding to the at least one blind sweep ultrasound scan. The at least one processor is further configured to estimate at least one fetal statistic based on the one or more anatomies and the gestational age.

[0008] In some embodiments, the at least one processor configured is further configured to: determine, using the first neural network, one or more bounding boxes for the one or more anatomies, and to calculate, based on the one or more bounding boxes, at least one measurement of the one or more anatomies. In some such embodiments, the at least one processor is further configured to: estimate the at least one fetal statistic based on applying the measurement and the gestational age to a lookup table.

[0009] In some embodiments, the at least one processor is further configured to: determine, using the first neural network, that the data corresponding to the at least one blind sweep ultrasound scan comprises a standard plane. In some such embodiments, the at least one processor is further configured to: calculate a measurement of the one or more anatomies present in the standard plane, and estimate the at least one fetal statistic based on applying the measurement and the gestational age to a lookup table.

[0010] In some embodiments, the at least one fetal statistic comprises one or more of a percentile fetal size, a small for gestational age (SGA) indication, and a large for gestational age (LGA) indication.

[0011] In some embodiments, the at least one processor is further configured to output the fetal statistic to the display

[0012] In some embodiments, the measurement of at least one of the one or more anatomies comprises at least one of a head circumference measurement, an abdominal circumference measurement, a femur length, or a biparietal diameter of a head.2024PF002873BRIEF DESCRIPTION OF THE DRAWINGS

[0013] FIG. 1 illustrates a block diagram of an ultrasound imaging system arranged in accordance with principles of the present disclosure.

[0014] FIG. 2 illustrates a first estimation protocol for the estimation of a statistic of interest of a fetus when using blind sweep ultrasound scanning techniques according to embodiments herein.

[0015] FIG. 3 illustrates a second estimation protocol for the estimation of a statistic of interest of a fetus when using blind sweep ultrasound scanning techniques according to embodiments herein.

[0016] FIG. 4 illustrates an example image from a blind sweep ultrasound scan wherein the anatomy is identified and a bounding box of the anatomy is determined according to embodiments herein.

[0017] FIG. 5 illustrates an example image from a standard plane from a blind sweep ultrasound scan according to embodiments herein.

[0018] FIG. 6 illustrates a computer-implemented method of detecting fetal growth according to embodiments herein.DETAILED DESCRIPTION

[0019] The following description of certain examples is illustrative in nature and is in no way intended to limit the disclosed technology or its applications or uses. In the following detailed description of examples of the present apparatuses, systems, and methods, reference is made to the accompanying drawings, which form a part hereof, and in which are shown by way of illustration specific examples in which the described apparatuses, systems, and methods may be practiced. These examples are described in sufficient detail to enable those skilled in the art to practice the presently disclosed apparatuses, systems, and methods, and it is to be understood that other examples may be utilized and that structural and logical changes can be made without departing from the spirit and scope of the present disclosure. Moreover, for the purpose of clarity, detailed descriptions of certain features will not be discussed when they would be apparent to those with skill in the art, so as not to obscure the description of the present technology. The following detailed description is therefore not to be taken in a limiting sense, and the scope of the present technology is defined only by the appended claims.

[0020] Users of medical imaging systems, such as ultrasound imaging systems, face technical challenges related to acquiring images using standard clinical protocols. For example, it may take two or more years to train sonographers to operate ultrasound imaging systems. The sonographer must be able to recognize anatomy in images and manipulate the probe in order to obtain images from standard planes, which vary based on the type of exam. The sonographer may also be required to take various measurements either from images (e.g., left ventricle internal diameter from a standard plane image) or other ultrasound data (e.g., blood flow rate from Doppler or m-mode data). Recently, some ultrasound systems are capable of providing support for sonographers or clinicians with less familiarity with2024PF002874 acquiring ultrasound data. For example, a user can perform guided sweeps, in which the user follows guidance provided via a device or system to reach a position or orientation for capturing an image of a standard plane. As a result, the less experienced user may be able to acquire various measurements used for health monitoring or diagnosis.

[0021] In prenatal care settings, fetal growth may be detected and monitored to determine if the fetus is developing normally. The user may acquire standard measurements of a fetus such as a head circumference measurement, an abdominal circumference measurements, a femur length measurement, and a biparietal diameter measurement of the head. Then, a combination of the acquired measurements or only the abdominal circumference measurement alone may be compared against, for example, a Hadlock table, including different percentile ranges for fetus measurements for a particular gestational age to estimate the fetal size and, in some cases, detect abnormalities in the fetus' growth or size. For example, the fetus may be abnormally small or large for its gestational age.

[0022] In some other examples, ultrasound imaging data can be acquired (e.g., by a novice user) by moving an ultrasound probe according to a predetermined grid (termed “blind sweeps”) over a subject (e.g., a pregnant subject’s abdomen). Such blind sweep ultrasound scanning protocols are useful, for example, in resource-constrained settings or in environments where experienced or trained users of ultrasound imaging systems may not be available. Additionally, blind sweep ultrasound scanning protocols may be useful for acquiring ultrasound imaging data for further processing by one or more machine learning models or other analysis algorithms.

[0023] Blind sweeps may result in a large set of ultrasound data, which may include two-dimensional and / or three-dimensional (volume) data. In some cases, only a subset of the data includes clinically relevant data. For example, standard planes for an exam (e.g., fetal exam, cardiac exam) may make up only a handful of planes of tens, hundreds, or even thousands of planes acquired by the blind sweep. Locating the standard planes in the blind sweep data may be difficult and / or time consuming.

[0024] However, blind sweeps do not always result in the novice user obtaining data from which standard planes can be extracted. As a result, blind sweep ultrasound scans may not be able to be used with existing clinical techniques (e.g., Hadlock tables) for estimating the percentile size of a fetus and other fetal statistics of interest (e.g., SGA and LGA).

[0025] The present disclosure describes a system and related methods for estimating a percentile fetal size (and corresponding fetal statistics of interest such as SGA and / or LGA) from ultrasound scans obtained by performing blind sweeps through the use of, for example, a neural network (NN) (or analysis algorithms of the sort such as artificial intelligence (Al) models and machine learning models). In some embodiments, the data corresponding to blind sweep ultrasound scans may be used as an input to a first neural network which may detect / identify one or more anatomies in the blind sweep ultrasound scan data. The first neural network may identify the one or more anatomies by analyzing2024PF002875 each frame (e.g., each image of a set of images) of the blind sweep ultrasound scan. It should be understood that anatomies in a blind sweep ultrasound scan data may include, for example, a head, an abdomen, a femur, a heart, and any other anatomy of interest in the blind sweep ultrasound scan.

[0026] In some cases, a bounding box may be applied to each of the identified anatomies (i.e., bounding / encompassing the identified anatomy) and, a statistic of interest of the number of bounding boxes may be calculated (e.g., a mean, a median, and / or a mode of the number of bounding boxes). In some other cases, a bounding box may be applied to each of the identified anatomies and a measurement may be performed on each of the bounding boxes (e.g., a size / length of the bounding box). A mean, a median, or a mode of the bounding box measurements may be taken if more than one bounding box measurement is performed on a particular anatomy, for example when it is found in multiple frames. It should be understood that such calculations and bounding box application may be performed by the first neural network. If no anatomies (or not all anatomies required for fetal growth percentile estimation) are detected, the technology may prompt the user to repeat the blind sweep protocol scan, or alert the user that the resulting data corresponding to the blind sweep protocol does not include the necessary information to make a fetal growth percentile estimation. The prompt and / or alert may be via displayed text, audio signal, and / or other indication.

[0027] In yet some other cases, the first neural network may determine if the data corresponding to the blind sweep ultrasound scan data includes a standard plane (e.g., an image acquired at a standard plane). Determining whether the blind sweep ultrasound scan data may include detecting fetal anatomies. If a standard plane is detected within the blind sweep ultrasound scan data, standard plane measurement(s) of the anatomies within the standard plane of the blind sweep data may be performed by the first neural network. In some embodiments, bounding boxes may be calculated around the anatomy (or anatomies) in the standard plane, and the measurements of the bounding box may be the measurement of the anatomy, similar to the non-standard plane case. If a standard plane is not detected, the first neural network, or the system as a whole, may output that no standard plane is detected and thus no estimation of percentile fetus size is made (i.e., no SGA or LGA estimation is made).

[0028] Additionally, the blind sweep ultrasound scan data may be used as an input to a second neural network (differing from the first neural network discussed herein) to determine a gestational age of the fetus in the blind sweep ultrasound scan. In some cases, this may be completed by the neural network using the entirety of the blind sweep ultrasound scan data as an input to determine the gestational age. The determined gestational age of the fetus may be in, for example, a number of days or a number of weeks. In some instances, the gestational age may be provided by the patient or by the user from either past ultrasound scanning techniques or other procedures for determining a gestational age, and thus, in such instances, the second neural network may not be utilized. In some embodiments, the second neural network may be robust to changes in fetal growth sizes and / or to fetal growth abnormalities when2024PF002876 predicting the gestational age of the fetus. In some examples, the second neural network may be an end- to-end convolutional neural network with attention -based architecture.

[0029] The first and second neural networks can be trained using training datasets comprising ultrasound data acquired by blind sweeps and / or other techniques. The training datasets may further include labels, annotations, or other data, for example standard plane captured anatomies. For example, the training dataset for the first neural network may include images with anatomy of interest labelled (e.g., heart, skull, femur) and bounding boxes and / or standard of care measurements around said anatomy of interest. In examples where standard frames are detected, images may include annotations as to whether or not the image is a standard frame, and if so, which standard frame (if there are multiple standard frames for an exam). The training data set for the second neural network may include many images from blind sweep scans or entire blind sweep scans (e.g., all of the images acquired during the blind sweep scan) labelled with the gestational age. Once the neural networks have been trained, the model receives data comprising blind sweep data, and the neural networks output gestational age, anatomical measurements, percentile of fetal size, fetal statistic of interest, or a combination thereof.

[0030] The first or second neural network may include one or more input nodes that receive training datasets (i.e., input layers). The input nodes can correspond to functions that receive the input and produce results. These results can be provided to one or more levels of intermediate nodes that each produce further results based on a combination of lower-level node results. A weighting factor can be applied to the output of each node before the result is passed to the next layer node. At a final layer, (“the output layer,”) one or more nodes can produce a value classifying the input that, once the model is trained, can be used to evaluate new data (e.g., gestational age). In some implementations, such neural networks, known as deep neural networks, can have multiple layers of intermediate nodes with different configurations, can be a combination of networks or machine learning / AI models that receive different parts of the input and / or input from other parts of the deep neural network, or are convolutions — partially using output from previous iterations of applying the neural network as further input to produce results for the current input.

[0031] A neural network can be trained with supervised or unsupervised learning. Testing data can then be provided to the neural network to assess accuracy. Testing data can be, for example, a portion of the entire dataset (e.g., 10%) held back to use for evaluation of the model. Output from the neural network can be compared to the desired or expected output for the training data and, based on the comparison, the neural network can be modified, such as by changing weights between nodes of the neural network and / or parameters of the functions used at each node in the neural network (e.g., applying a loss function). Based on the results of the neural network evaluation, and after applying the described modifications, the neural network can then be retrained to evaluate new data.2024PF002877

[0032] It should be understood that use of “first neural network” and “second neural network” is for example and differentiation between the neural networks only. For example, a first neural network may be used to determine the gestational age while a second neural network may be used to identify the anatomy. In some other examples, multiple neural networks may be used to determine an anatomy in the blind sweep ultrasound scan data or to determine the gestational age.

[0033] In some embodiments, the measurements acquired from the identified one or more anatomies, the determined gestational age, or a combination thereof may be compared to a lookup table which estimates a fetal statistic of interest (e.g., percentile of the fetal size of the fetus in the blind sweep ultrasound scan, if the fetus is LGA or SGA, if the fetus has an average gestational age, if the fetus is at a healthy size for the gestational age). In some cases, the lookup table may provide specific statistics tailored for a particular population (i.e., per country, per state, per region) for estimating a percentile of the fetal size. Comparing the measurements acquired from the identified one or more anatomies to a lookup table may be understood as applying to a lookup table or passing through a lookup table.

[0034] In some other embodiments, if a standard plane is determined to be in the blind sweep ultrasound scan data, the measurements obtained from the standard plane may be applied, in combination with the determined gestational age, to the lookup table.

[0035] Such lookup tables may be a Hadlock table, an integral lookup table, a World Health Organization (WHO) lookup table, (or any lookup table of the sort). Alternatively, in some embodiments, a lookup table may be developed for such applications, where the lookup table may be developed based on various amounts of previously acquired ground truth data where a percentile fetal size is confirmed for each gestational age and each corresponding anatomy identification by using accepted clinical procedures.

[0036] In some implementations, the estimated fetal statistics (e.g., percentile fetus size, SGA, LGA) may be used to detect abnormalities in the growth of a fetus. For example, if the estimated percentile of fetus size of the fetus is outside the parameters for a normal percentile of fetus size, it may be recommended to perform additional non-blind sweep ultrasound scans of the fetus.

[0037] In some other implementations, the estimated fetal statistics (e.g., percentile fetus size, SGA, LGA) may be outputted to the user and / or patient with additional information on what anatomy the fetal statistic was estimated based on. For example, an SGA determination may be outputted to the user / patient with included information clarifying that the determination was made based on a femur length.

[0038] Advantages of the disclosed technology may include introducing system(s) and method(s) to estimate a fetus size percentile while using blind sweep ultrasound scanning techniques, as such blind sweep ultrasound scanning techniques, currently, are often unable to estimate the fetus size percentile. Additionally, the disclosed technology may, when using blind sweep ultrasound scanning techniques,2024PF002878 assist in detecting abnormalities in the growth of a fetus, thus allowing for proper steps to be taken to ensure the health of the fetus. For example, if an abnormality is detected (i.e., SGA or LGA), it may be recommended to undergo further ultrasound scanning of the fetus or to consult a clinical practitioner or medical expert of the sort.

[0039] Additionally, in current clinical practices, if a patient does not have previous ultrasound scan data, gestational age data, or fetal data, an accurate fetal size estimation / determination may not be made. However, technology disclosed herein enables the estimation of fetal size, and thus enables the determination of abnormality of fetal growth (i.e., SGA or LGA) with a single blind sweep ultrasound scan, with no previous fetal data.

[0040] It should be understood that, while examples herein relate to estimating a fetus size percentile of a fetus while using blind sweep ultrasound scanning techniques, it should be appreciated that the disclosed technology may be used to estimate other statistics of interest of the fetus while using blind sweep ultrasound scanning techniques.

[0041] FIG. 1 illustrates a block diagram of an ultrasound imaging system 100 arranged in accordance with principles of the present disclosure. In the ultrasound imaging system 100 of FIG. 1, an ultrasound probe 112 includes a transducer array 114 for transmitting ultrasonic waves and receiving echo information. The transducer array 114 can be implemented as a linear array, convex array, a phased array, and / or a combination thereof. The transducer array 114, for example, can include a two- dimensional array (as shown) of transducer elements capable of scanning in both elevation and azimuth dimensions for 2D and / or 3D imaging. The transducer array 114 can be coupled to a microbeamformer 116 in the probe 112, which controls transmission and reception of signals by the transducer elements in the array. In this example, the microbeamformer 116 is coupled by the probe cable to a transmit / receive (T / R) switch 118, which switches between transmission and reception and protects the main beamformer 122 from high-energy transmit signals. In some embodiments, the T / R switch 118 and other elements in the system can be included in the ultrasound probe 112 rather than in a separate ultrasound system base. In some embodiments, the ultrasound probe 112 may be coupled to the ultrasound imaging system via a wireless connection (e.g., WiFi, Bluetooth).

[0042] The transmission of ultrasonic beams from the transducer array 114 under control of the microbeamformer 116 is directed by the transmit controller 120 coupled to the T / R switch 118 and the beamformer 122, which receives input from the user’s operation of the user interface (e.g., control panel, touch screen, console) 125. The user interface 125 may include soft and / or hard controls. One of the functions controlled by the transmit controller 120 is the direction in which beams are steered. Beams may be steered straight ahead from (orthogonal to) the transducer array 114, or at different angles for a wider field of view. The partially beamformed signals produced by the microbeamformer 116 are coupled via channels 115 to a main beamformer 122 where partially beamformed signals from2024PF002879 individual patches of transducer elements are combined into a fully beamformed signal. In some embodiments, microbeamformer 116 is omitted and the transducer array 114 is coupled via channels 115 to the beamformer 122. In some embodiments, the system 100 can be configured (e.g., include a sufficient number of channels 115 and have a transmi t / receive controller programmed to drive the transducer array 114) to acquire ultrasound data responsive to a plane wave or diverging beams of ultrasound transmitted toward the subject. In some embodiments, the number of channels 115 from the ultrasound probe may be less than the number of transducer elements of the transducer array 114 and the system can be operable to acquire ultrasound data packaged into a smaller number of channels than the number of transducer elements.

[0043] The beamformed signals are coupled to a signal processor 126. The signal processor 126 can process the received echo signals in various ways, such as bandpass filtering, decimation, I and Q component separation, and / or harmonic signal separation. The signal processor 126 can also perform additional signal enhancement such as speckle reduction, signal compounding, and noise elimination. The processed signals are coupled to a B-mode processor 128, which can employ amplitude detection for the imaging of structures in the body. The signals produced by the B-mode processor 128 are coupled to a scan converter 130 and a multiplanar reformatter 132. The scan converter 130 arranges the echo signals in the spatial relationship from which they were received in a desired image format. For instance, the scan converter 130 can arrange the echo signal into a two-dimensional (2D) sector-shaped format, or a pyramidal three-dimensional (3D) image. The multiplanar reformatter 132 can convert echoes, which are received from points in a common plane in a volumetric region of the body into an ultrasonic image of that plane, as described in U.S. Pat. No. 6,443,896 (Detmer).

[0044] A volume Tenderer 134 converts the echo signals of a 3D data set into a projected 3D image as viewed from a given reference point, e.g., as described in U.S. Pat. No. 6,530,885 (Entrekin et al.). In some implementations, the system 100 can additionally or alternatively be configured to perform 3D and / or 4D acquisitions, such as protocolized 3D / 4D acquisitions. The 2D or 3D images can be coupled from the scan converter 130, multiplanar reformatter 132, and volume Tenderer 134 to at least one processor 137 for further image processing operations. For example, the at least one processor 137 can include an image processor 136 configured to perform further enhancement and / or buffering and temporary storage of imaging data for display on an image display 138. The display 138 can include a display device implemented using a variety of display technologies, such as LCD, LED, OLED, or plasma display technology. The at least one processor 137 can include a graphics processor 140, which can generate graphic overlays for display with the ultrasound images. These graphic overlays can contain, e.g., standard identifying information such as patient name, date and time of the image, imaging parameters, and the like. For these purposes the graphics processor 140 receives input from the user interface 125, such as a typed patient name. The user interface 125 can also be coupled to the2024PF0028710 multiplanar reformatter 132 for selection and control of a display of multiple multiplanar reformatted (MPR) images.

[0045] The user interface 125 can include one or more mechanical controls, such as buttons, dials, a trackball, a physical keyboard, and others, which may also be referred to herein as hard controls. Alternatively or additionally, the user interface 125 can include one or more soft controls, such as buttons, menus, soft keyboard, and other user interface control elements implemented for example using touch-sensitive technology (e.g., resistive, capacitive, or optical touch screens). One or more of the user controls can be co-located on a control panel 124. For example, one or more of the mechanical controls can be provided on a console and / or one or more soft controls can be co-located on a touch screen, which can be attached to or integral with the console. The display 138 and the user interface 125 can be included in an I / O component, via which outputs are provided by the system 100 and / or inputs are received by the system 100.

[0046] In some implementations, the user interface 125 can receive inputs and provide outputs of the disclosed system. For example, the user interface 125 can receive a user input specifying an exam type. In some implementations the one or more outputs of the system can be provided via one or more graphical user interfaces (e.g., via the display 138).

[0047] The at least one processor 137 (e.g., the image processor 136, the graphics processor 140, or a different processor) can perform functions associated with acquiring and analyzing medical imaging data, as described herein. For example, the at least one processor may implement one or more neural networks or other machine learning or Al models for recognizing anatomy in images and / or images of standard planes.

[0048] Although described as separate processors, it will be understood that the functionality of any of the processors described herein can be implemented in a single processor (e.g., a CPU or GPU implementing the functionality of processor 137) or fewer number of processors than described in this example. In some embodiments, the at least one processor 137 can be hardware -based (e.g., include multiple layers of interconnected nodes implemented in hardware). In some embodiments, the at least one processor 137 can be implemented at other processing stages, e.g., prior to the processing performed by the image processor 136, volume Tenderer 134, multiplanar reformatter 132, and / or scan converter 130. In some embodiments, the at least one processor 137 can be implemented to process ultrasound data in the channel domain, beamspace domain (e.g., before or after beamformer 122), the IQ domain (e.g., before, after, or in conjunction with signal processor 126), and / or the k-space domain. As described, in some embodiments, functionality of two or more of the processing components (e.g., beamformer 122, signal processor 126, B-mode processor 128, scan converter 130, multiplanar reformatter 132, volume Tenderer 134, at least one processor 137, image processor 136, graphics processor 140, etc.) can be combined into a single processing unit and / or divided between multiple2024PF0028711 processing units. The processing units can be implemented in software, hardware, or a combination thereof. For example, the at least one processor 137 can include one or more graphical processing units (GPU). In another example, beamformer 122 can include an application specific integrated circuit (ASIC).

[0049] The at least one processor 137 can be coupled to one or more computer-readable media (e.g., memory 142) included in the system 100, which can be non-transitory. The one or more computer- readable media can carry instructions and / or a computer program that, when executed, cause the at least one processor 137 to perform operations described herein. A computer program can be stored / distributed on any suitable medium, such as an optical storage medium or a solid- state medium supplied together with or as part of other hardware, and can also be distributed in other forms, such as via the Internet or other wired or wireless telecommunication systems. Furthermore, embodiments can take the form of a computer program product accessible from a computer- readable medium providing program code for use by or in connection with a computer or any device or system that executes instructions. For the purposes of this disclosure, a computer-readable medium can generally be any tangible apparatus that can contain, store, communicate, propagate, and / or transport the program for use by or in connection with the instruction execution device. The computer- readable medium can be, for example, without limitation, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, and / or a propagation medium. Non-limiting examples of a computer readable medium include a semiconductor or solid state memory, magnetic tape, a removable computer diskette, a random access memory (RAM), a read-only memory (ROM), a rigid magnetic disk, and / or an optical disk. Optical disks can include compact disk read only memory (CD-ROM), compact disk-read / write (CD-R / W), and / or DVD.

[0050] The system 100 may be capable of performing the techniques disclosed herein according to principles of the present disclosure. However, in some instances, the system 100 may acquire the blind sweep data, and that data may be provided to a computing system with one or more processors (e.g., signal processor, image processor, graphics processor, multiplanar reformatter, volume Tenderer), a non-transitory computer readable medium, and a user interface. The computing system may perform some or all of the remaining non-image acquisition functions in these embodiments. Accordingly, while the examples provided herein describe data analysis functions being performed by an ultrasound imaging system, such as system 100, the disclosure is not limited to these embodiments.

[0051] FIG. 2 illustrates a first estimation protocol 200 for the estimation of a statistic of interest of a fetus when using blind sweep ultrasound scanning techniques according to embodiments herein.

[0052] The first estimation protocol 200 may be performed using the system 100 of FIG. 1. For example, the at least one processor 137 may take data corresponding to the blind sweep ultrasound scan2024PF0028712 as an input to perform the first estimation protocol 200 used for the estimation of a statistic of interest of a fetus in the blind sweep ultrasound scan.

[0053] In some embodiments, the first estimation protocol 200 begins with acquiring blind sweep data corresponding to a performed blind sweep ultrasound scanning technique at block 202. However, in other examples, the blind sweep may have already been performed, and the blind sweep data may be retrieved from a memory, such as memory 142. When the blind sweep data has already been acquired, the protocol 200 may be performed by a computing system that is provided or has access to the blind sweep data in some embodiments (e.g., a radiology workstation at a hospital).

[0054] The blind sweep data may be passed as an input to a first neural network (e.g., a fetal anatomy detector) at block 204 to detect anatomies in the blind sweep data. For example, each image in the blind sweep data may be analyzed to detect anatomies. Such identification may include determining a bounding box of one or more anatomies identified in the blind sweep data as performed at block 204. In some instances, measurements may be performed on the bounding boxes, thus measuring the identified anatomies in the blind sweep data. For example, a length and / or width of a bounding box around a fetal skull may be measured to provide a fetal skull length or diameter. When a fetal anatomy is detected in multiple images, a mean, median, or mode may be obtained from the measurements of the bounding boxes and used for fetal size estimation at block 206. For example, if a fetal femur is detected in multiple frames, the lengths of the femur measured in each of the frame may be averaged to obtain the final fetal measurement.

[0055] The blind sweep data may be passed as an input to a second neural network (e.g., a gestational age regressor) at block 208 to determine a gestational age of the fetus in the blind sweep data at block 210. In some cases, this may be completed by the neural network using the entirety of the blind sweep ultrasound scan data as an input to determine the gestational age. In some instances, the gestational age may be provided by the patient or by the user from either past ultrasound scanning techniques or other procedures for determining a gestational age, and thus, in such instances, the second neural network may not be utilized. In some examples, the second neural network may be a regressor model that may provide a relationship between the blind sweep data and determined gestational age (i.e., providing an end-to-end convolution).

[0056] The fetal measurements determined at block 206 and the determined gestational age from block 210 may be applied to a lookup table at block 212 to estimate a percentile of fetal size. In some embodiments, a lookup table may be developed for such applications, where the lookup table may be developed based on previously acquired robust ground truth data where a percentile fetal size is confirmed for each gestational age and corresponding anatomy identification.

[0057] Based on the lookup table, a statistic of interest of a fetus, such as a percentile output of fetal size may be output at block 214. The output may be provided to a display such as display 138. In some2024PF0028713 embodiments, the estimated fetal gestational age output at block 210 may also be provided to the display. In some embodiments, the percentile output of fetal size and / or gestational age may be stored in a memory, such as memory 142.

[0058] In some situations, such as when the blind sweep data is incomplete and / or poor quality, the neural networks may not be able to identify all of the anatomies or other features needed to calculate fetal size and / or gestational age. The inability to identify features may be based on a failure to detect a feature at all or a feature is detected, but a confidence score of the neural network for the feature is below a threshold value (e.g., 50%, 70%, 90%). In these cases, instead of a statistic of interest, an error message may be output at block 214.

[0059] FIG. 3 illustrates a second estimation protocol 300 for the estimation of a statistic of interest of a fetus when using blind sweep ultrasound scanning techniques according to embodiments herein.

[0060] The second estimation protocol 300 may be performed using the system 100 of FIG. 1. For example, the at least one processor 137 may take data corresponding to the blind sweep ultrasound scan as an input to perform the second estimation protocol 300 used for the estimation of a statistic of interest of a fetus in the blind sweep ultrasound scan.

[0061] In some embodiments, the second estimation protocol 300 begins with acquiring blind sweep data corresponding to a performed blind sweep ultrasound scanning technique at block 302. However, in other examples, the blind sweep may have already been performed, and the blind sweep data may be retrieved from a memory, such as memory 142. When the blind sweep data has already been acquired, the protocol 300 may be performed by a computing system that is provided or has access to the blind sweep data in some embodiments.

[0062] The blind sweep data may be passed as an input to a first neural network (e.g., a fetal anatomies detector) at block 304 to detect if one or more standard planes (e.g., image from a standard plane) is included in the acquired blind sweep ultrasound scan data. Identified anatomies may include, for example, the head of the fetus, the abdomen of the fetus, the femur of the fetus, and / or any other anatomy of interest in the standard plane(s). In contrast to the embodiment described in FIG. 2, the neural network may be trained to not only detect fetal anatomies but also the arrangement and view of the fetal anatomies within frames to determine the image is from a standard plane. Once one or more standard planes are identified, the neural network may perform measurements on the identified anatomies. In some embodiments, the neural network may use the same or similar techniques as described with reference to FIG. 2. For example, the neural network may apply bounding boxes to the identified anatomies and make measurements of the bounding boxes.

[0063] The blind sweep data may be passed as an input to a second neural network (e.g., a gestational age regressor) at block 308 to determine and output a gestational age of the fetus in the blind sweep2024PF0028714 data at block 310. The second neural network may be substantially the same as the second neural network described with reference to FIG. 2 in some embodiments.

[0064] The measurements of the identified anatomies from the standard plane(s) and the determined gestational age may be applied to a lookup table at block 312 to estimate a fetal size percentile (e.g., if the fetus is SGA, LGA, or normal), where the lookup table may include, for example, a Hadlock table, an integral lookup table, a WHO look up table, a lookup table of the sort used in current clinical practice. In some embodiments, a lookup table may be developed for such applications, where the lookup table may be developed based on previously acquired robust ground truth data where a percentile fetal size is confirmed for each gestational age and corresponding anatomy identification. Accordingly, abnormal fetus growth may be detected based on if the fetus is SGA or LGA for its gestational age.

[0065] Based on the lookup table, a statistic of interest of a fetus, such as whether the fetus is small for gestational age, large for gestational age, or normal for gestational age may be output at block 314. The output may be provided to a display such as display 138. In some embodiments, the estimated fetal gestational age output at block 310 may also be provided to the display. In some embodiments, the statistic of interest of the fetus and / or gestational age may be stored in a memory, such as memory 142.

[0066] In some situations, such as when the blind sweep data is incomplete and / or poor quality, the neural networks may not be able to identify all of the standard planes needed to calculate fetal size and / or features needed to calculate gestational age. The inability to identify standard planes or features may be based on a failure to detect a plane or feature at all or a plane or feature is detected, but a confidence score of the neural network for the feature is below a threshold value (e.g., 50%, 70%, 90%). In these cases, instead of a statistic of interest, an error message may be output at block 314. In some embodiments, if gestational age can be calculated, but one or more standard planes could not be determined, the system may suggest to a user to perform the protocol described in FIG. 2.

[0067] In some cases, the protocols described in FIGS. 2 and 3 may provide similar information regarding fetal growth. In some settings, clinicians may be more comfortable with the protocol described with reference to FIG. 3 because it utilizes standard planes, measurements and look-up tables familiar to the clinician from performing manual scans and analysis. However, the analysis, measurements, and / or outputs described with reference to FIG. 3 may not necessarily be better quality / accuracy than those described with reference to FIG. 2.

[0068] FIG. 4 illustrates an example of an image 400 from a blind sweep ultrasound scan wherein the anatomy 404 is identified and a bounding box 402 of the anatomy 404 is determined according to embodiments herein.

[0069] In some embodiments, the first neural network (e.g., the neural network shown at 204) may be used to identify an anatomy 404 in an image 400 from a blind sweep ultrasound scan. In the example2024PF0028715 shown in FIG. 4, the anatomical feature identified is a fetal abdomen in a non-standard plane. A bounding box 402 surrounding / encompassing the anatomy 404 may be determined by the neural network. The bounding box 402 may be measured to obtain measurements of the anatomy 404. In the example shown in FIG. 4, a diameter or circumference of the abdomen may be determined from the dimensions of the bounding box 402. The measurement may be used in a lookup table, in combination with the determined gestational age, to provide the statistic of interest of the fetus, for example as described with reference to FIG. 2, a percentile output of fetal size may be provided.

[0070] FIG. 5 illustrates an example image of a standard plane 500 from a blind sweep ultrasound scan according to embodiments herein.

[0071] As in some embodiments, the first neural network (e.g., neural network shown at 304) may determine if a standard plane image 500 is included in the blind sweep ultrasound scan. Determination of a standard plane may include determining if an anatomy 502 is present in the image 500. In the example shown in FIG. 5, the image 500 is a standard plane of a fetal skull. The neural network may make a measurement on the anatomy 502. Similar to FIG. 4, a bounding box 504 may be fitted to the anatomy 502. In the example shown in FIG. 5, measurements of the bounding box 502 may be used to obtain length, width, and / or circumference of the skull. In some embodiments, alternatively or in addition to the bounding box 502, an ellipse 506 may be fitted to the anatomy 502, and measurements from the ellipse may be acquired. The measurements may be used in a look up table, in combination with the determined gestational age, to provide a statistic of interest of the fetus. For example, an indication of percentile of fetal size, SGA, LGA, or normal fetus size may be provided.

[0072] FIG. 6 illustrates a computer-implemented method 600 of detecting fetal growth, according to embodiments herein. The computer-implemented method 600 may be implemented in whole or in part by an ultrasound imaging system, such as system 100 in some examples. In other examples, the computer-implemented method 600 may be implemented in whole or in part by a computing system, such as a radiology workstation, that may not have ultrasound imaging capabilities. The illustrated method 600 includes receiving 602 data corresponding to at least one blind sweep ultrasound scan. For example, the data corresponding to the at least one blind sweep ultrasound scan may be received by the at least one processor 137 for use in block 202 and / or block 302. In some examples, the data corresponding to the at least one blind sweep ultrasound scan may be stored in memory 142, and the processor 137 may receive the data from the memory 142. In other examples, the data corresponding to the at least one blind sweep ultrasound scan may have been acquired by an ultrasound imaging system, such as system 100, and the data corresponding to the at least one blind sweep ultrasound scan may be provided to a memory and / or at least one processor of a computing system separate from the ultrasound imaging system.2024PF0028716

[0073] The method 600 further includes identifying 604, using a first neural network, one or more anatomies in the data corresponding to the at least one blind sweep ultrasound scan. For example, the neural network at block 204 or neural network at block 304 may be implemented by the image processor 136 and / or graphics processor 140 to identify the one or more anatomies.

[0074] The method 600 further includes determining 606, using a second neural network, a gestational age based on the data corresponding to the at least one blind sweep ultrasound scan. For example, the neural network at block 208, or the neural network at block 308 may be implemented by the image processor 136 and / or graphics processor 140 to determine the gestational age as illustrated at block 210 and block 310 respectively.

[0075] The method 600 further includes estimating 608 at least one fetal statistic based on the one or more anatomies and the gestational age. For example, at block 214 and / or block 314 a fetal statistic is estimated based on the one or more anatomies determined in block 204 and block 304 and the gestational age determined in block 208 and block 308.

[0076] In some embodiments of the method 600, identifying the one or more anatomies further comprises: determining, using the first neural network, one or more bounding boxes for the one or more anatomies, and calculating, based on the one or more bounding boxes, at least measurement of the one or more anatomies. For example, at block 206, the bounding box may be determined and used for calculating a measurement of the one or more anatomies. In some such embodiments, estimating the at least one fetal statistic further comprises: estimating the at least one fetal statistic based on applying the at least one measurement and the gestational age to a lookup table. For example, at block 212 a lookup table may be used for the estimation of the at least one fetal statistic.

[0077] In some embodiments of the method 600, identifying the one or more anatomies further comprises: determining, using the first neural network, that the data corresponding to the at least one blind sweep ultrasound scan comprises a standard plane. For example, at block 306 a standard plane may be detected in the blind sweep ultrasound scan data. In some such embodiments, estimating the at least one fetal statistic further comprises: calculating a measurement of at least one of the one or more anatomies present in the standard plane, and estimating the at least one fetal statistic based on applying the identified one or more anatomies and the gestational age to a lookup table. For example, at block 312, a calculation may be performed on the standard plane and the fetal statistic may be estimated based on the calculated measurement and the determined gestational age. In some such embodiments, the measurement of at least one of the one or more anatomies comprise at least one of a head circumference measurement, an abdominal circumference measurement, a femur length, or a biparietal diameter of a head. For example, at block 312, the calculation may be based on said anatomies.

[0078] In some embodiments of the method 600, the at least one fetal statistic comprises one or more of a percentile fetal size, a small for gestational age (SGA) indication, and a large for gestational age2024PF0028717(LGA) indication. For example, at block 214 and block 314 the estimated fetal statistic comprises said indications.

[0079] The systems, methods, and techniques disclosed herein may improve the ability for data from blind sweeps to be used for detecting and monitoring fetal growth. This may allow expanded use of ultrasound imaging for maternal and fetal healthcare in areas where there are few or no trained sonographers available.

[0080] In various examples where components, systems and / or methods are implemented using a programmable device, such as a computer-based system or programmable logic, it should be appreciated that the above-described systems and methods can be implemented using any of various known or later developed programming languages, such as “Python”, “C”, “C++”, “FORTRAN”, “Pascal”, “VHDL” and the like. Accordingly, various storage media, such as magnetic computer disks, optical disks, electronic memories and the like, can be prepared that can contain information that can direct a device, such as a computer, to implement the above-described systems and / or methods. Once an appropriate device has access to the information and programs contained on the storage media, the storage media can provide the information and programs to the device, thus enabling the device to perform functions of the systems and / or methods described herein. For example, if a computer disk containing appropriate materials, such as a source file, an object file, an executable file or the like, were provided to a computer, the computer could receive the information, appropriately configure itself and perform the functions of the various systems and methods outlined in the diagrams and flowcharts above to implement the various functions. That is, the computer could receive various portions of information from the disk relating to different elements of the above-described systems and / or methods, implement the individual systems and / or methods and coordinate the functions of the individual systems and / or methods described above.

[0081] In view of this disclosure it is noted that the various methods and devices described herein can be implemented in hardware, software, and / or firmware. Further, the various methods and parameters are included by way of example only and not in any limiting sense. In view of this disclosure, those of ordinary skill in the art can implement the present teachings in determining their own techniques and needed equipment to affect these techniques, while remaining within the scope of the invention. The functionality of one or more of the processors described herein may be incorporated into a fewer number or a single processing unit (e.g., a CPU) and may be implemented using application specific integrated circuits (ASICs) or general -purpose processing circuits which are programmed responsive to executable instructions to perform the functions described herein.

[0082] Although the present system may have been described with particular reference to an ultrasound imaging system, it is also envisioned that the present system can be extended to other medical imaging systems where one or more images are obtained in a systematic manner. Accordingly,2024PF0028718 the present system may be used to obtain and / or record image information related to, but not limited to renal, testicular, breast, ovarian, uterine, thyroid, hepatic, lung, musculoskeletal, splenic, cardiac, arterial and vascular systems, as well as other imaging applications related to ultrasound-guided interventions. Further, the present system may also include one or more programs which may be used with conventional imaging systems so that they may provide features and advantages of the present system. Certain additional advantages and features of this disclosure may be apparent to those skilled in the art upon studying the disclosure, or may be experienced by persons employing the novel system and method of the present disclosure. Another advantage of the present systems and method may be that conventional medical image systems can be easily upgraded to incorporate the features and advantages of the present systems, devices, and methods.

[0083] Of course, it is to be appreciated that any one of the examples, examples or processes described herein may be combined with one or more other examples, examples and / or processes or be separated and / or performed amongst separate devices or device portions in accordance with the present systems, devices and methods.

[0084] Finally, the above-discussion is intended to be merely illustrative of the present systems and methods and should not be construed as limiting the appended claims to any particular example or group of examples. Thus, while the present system has been described in particular detail with reference to exemplary examples, it should also be appreciated that numerous modifications and alternative examples may be devised by those having ordinary skill in the art without departing from the broader and intended spirit and scope of the present systems and methods as set forth in the claims that follow. Accordingly, the specification and drawings are to be regarded in an illustrative manner and are not intended to limit the scope of the appended claims.

Claims

2024PF0028719CLAIMS1. An ultrasound imaging system comprising: at least one processor configured to: receive data corresponding to at least one blind sweep ultrasound scan; identify one or more anatomies in the data corresponding to the at least one blind sweep ultrasound scan; determine a gestational age based on the data corresponding to the at least one blind sweep ultrasound scan; and estimate at least one fetal statistic based on the one or more anatomies and the gestational age.

2. The ultrasound imaging system of claim 1, wherein the at least one processor is further configured to: determine one or more bounding boxes for the one or more anatomies; and calculate, based on the one or more bounding boxes, at least one measurement of the one or more anatomies.

3. The ultrasound imaging system of claim 2, wherein the at least one processor is further configured to: estimate the at least one fetal statistic based on applying the measurement and the gestational age to a lookup table.

4. The ultrasound imaging system of claim 1, 2 or 3, wherein the at least one processor is further configured to: determine that the data corresponding to the at least one blind sweep ultrasound scan comprises a standard plane.

5. The ultrasound imaging system of claim 4, wherein the at least one processor is further configured to: calculate a measurement of at least one of the one or more anatomies present in the standard plane; and estimate the at least one fetal statistic based on applying the measurement and the gestational age to a lookup table.2024PF00287206. The ultrasound imaging system of any of the preceding claims, wherein the at least one fetal statistic comprises one or more of a percentile fetal size, a small for gestational age (SGA) indication, and a large for gestational age (LGA) indication.

7. The ultrasound imaging system of any of the preceding claims, further comprising a display, wherein the at least one processor is further configured to output the fetal statistic to the display.

8. The ultrasound imaging system of claim 2, wherein the measurement of at least one of the one or more anatomies comprises at least one of a head circumference measurement, an abdominal circumference measurement, a femur length, or a biparietal diameter of a head.

9. The ultrasound imaging system of any of the preceding claims, wherein the at least one processor configured to: identify, using a first neural network, the one or more anatomies in the data corresponding to the at least one blind sweep ultrasound scan; and determine, using a second neural network, the gestational age based on the data corresponding to the at least one blind sweep ultrasound scan.

10. The ultrasound imaging system of claim 9 when dependent on claim 2, wherein the at least one processor configured to: determine, using the first neural network, the one or more bounding boxes for the one or more anatomies.

11. The ultrasound imaging system of claim 9 when dependent on claim 4, wherein the at least one processor is configured to: determine, using the first neural network, that the data corresponding to the at least one blind sweep ultrasound scan comprises the standard plane12. A computer-implemented method of estimating at least one fetal statistic, the method comprising: receiving data corresponding to at least one blind sweep ultrasound scan; identifying one or more anatomies in the data corresponding to the at least one blind sweep ultrasound scan; determining a gestational age based on the data corresponding to the at least one blind sweep ultrasound scan; and estimating the at least one fetal statistic based on the one or more anatomies and the gestational age.2024PF002872113. The computer-implemented method of claim 12, wherein identifying the one or more anatomies further comprises: determining one or more bounding boxes for the one or more anatomies; and calculating, based on the one or more bounding boxes, at least one measurement of the one or more anatomies.

14. The computer- implemented method of claim 13, wherein estimating the at least one fetal statistic further comprises: estimating the at least one fetal statistic based on applying the at least one measurement and the gestational age to a lookup table.

15. The computer-implemented method of claim 12, 13 or 14, wherein identifying the one or more anatomies further comprises: determining that the data corresponding to the at least one blind sweep ultrasound scan comprises a standard plane.

16. The computer- implemented method of claim 15, wherein estimating the at least one fetal statistic further comprises: calculating a measurement of at least one of the one or more anatomies present in the standard plane; and estimating the at least one fetal statistic based on applying the identified one or more anatomies and the gestational age to a lookup table.

17. The computer-implemented method of any of claims 12 to 16, wherein the at least one fetal statistic comprises one or more of a percentile fetal size, a small for gestational age (SGA) indication, and a large for gestational age (LGA) indication.

18. The computer-implemented method of claim 16, wherein the measurement of at least one of the one or more anatomies comprise at least one of a head circumference measurement, an abdominal circumference measurement, a femur length, or a biparietal diameter of a head.

19. The computer-implemented method of any of claims 12 to 18, wherein identifying the one or more anatomies in the data corresponding to the at least one blind sweep ultrasound scan comprises using a first neural network; and wherein determining the gestational age based on the data corresponding to the at least one blind sweep ultrasound scan comprises using a second neural network.

20. The computer-implemented method of claim 19 when dependent on claim 13,2024PF0028722 wherein determining the one or more bounding boxes for the one or more anatomies comprises using the first neural network.

21. The computer-implemented method of claim 19 when dependent on claim 15, wherein determining that the data corresponding to the at least one blind sweep ultrasound scan comprises the standard plane comprises using the first neural network.

22. A computer program product comprising instructions that, when executed by a computing system, cause the computing system to carry out the method of any one of claims 12 to 21.

Images