System and method for ultrasound image quality determination

Abstract

The present invention is entitled a system and method for ultrasound image quality determination. The present invention provides a method and system for evaluating the image quality of an ultrasound image. In one example, the method includes determining a probe position quality parameter of an ultrasound image, the probe position quality parameter representing the quality level of the ultrasound image relative to the position of an ultrasound probe used to acquire the ultrasound image; determining one or more acquisition setting quality parameters of the ultrasound image, each acquisition setting quality parameter representing the corresponding quality level of the ultrasound image relative to the corresponding acquisition setting used to acquire the ultrasound image; and providing feedback to a user of the ultrasound system based on the probe position quality parameter and / or the one or more acquisition setting quality parameters, the probe position quality parameter and each acquisition setting quality parameter being determined based on outputs from a separate image quality assessment model.

Description

Technical Field

[0001] Embodiments of the subject matter disclosed herein relate to ultrasound imaging, and more particularly, to improving image quality for ultrasound imaging. Background Art

[0002] Medical ultrasound is an imaging modality that uses ultrasound waves to detect the internal structures of a patient's body and produce corresponding images. For example, an ultrasound probe including multiple transducer elements transmits ultrasound pulses, which are reflected or transmitted back, refracted, or absorbed by structures in the body. The ultrasound probe then receives the reflected echoes, which are processed into an image. Ultrasound images of internal structures can be saved for later analysis by clinicians to aid in diagnosis and / or can be displayed on a display device in real time or near real time. Summary of the invention

[0003] In one embodiment, a method for an ultrasound system includes determining a probe position quality parameter of an ultrasound image acquired with the ultrasound system, the probe position quality parameter representing a quality level of the ultrasound image relative to a position of an ultrasound probe used to acquire the ultrasound image; determining one or more acquisition setting quality parameters of the ultrasound image, each acquisition setting quality parameter representing a corresponding quality level of the ultrasound image relative to a corresponding acquisition setting used to acquire the ultrasound image; and providing feedback to a user of the ultrasound system based on the probe position quality parameter and / or one or more acquisition setting quality parameters, the probe position quality parameter and each acquisition setting quality parameter being determined based on output from a separate image quality assessment model.

[0004] The above advantages and other advantages and features of the present specification will be apparent from the following detailed description when taken alone or in conjunction with the accompanying drawings. It should be understood that the above summary is provided to introduce a selection of concepts further described in the detailed description in a simplified form. This is not meant to identify key or essential features of the claimed subject matter, the scope of which is uniquely defined by the claims that follow the detailed description. In addition, the claimed subject matter is not limited to embodiments that address any disadvantages mentioned above or in any part of this disclosure. BRIEF DESCRIPTION OF THE DRAWINGS

[0005] Various aspects of the present disclosure may be better understood by reading the following detailed description and referring to the accompanying drawings, in which:

[0006] Figure 1 A block diagram of an exemplary embodiment of an ultrasound system is shown;

[0007] Figure 2 is a schematic diagram illustrating a system for generating an ultrasound image under optimized parameter settings according to an exemplary embodiment;

[0008] Figure 3 is a schematic diagram illustrating an exemplary ultrasound image quality problem;

[0009] Figure 4 is a flow chart illustrating an exemplary method for determining ultrasound scan quality according to one embodiment;

[0010] Figure 5 is a flow chart illustrating an exemplary method for scanning a setup model;

[0011] Figure 6 is a schematic diagram showing exemplary inputs and results of a scan setup model; and

[0012] Figures 7 to 10 is an exemplary graphical user interface that includes notifications that may be generated and output for display on a display device. DETAILED DESCRIPTION

[0013] Medical ultrasound imaging generally includes placing an ultrasound probe including one or more transducer elements onto an imaging subject (such as a patient) at the location of a target anatomical feature (e.g., abdomen, chest, etc.). The image is acquired by the ultrasound probe and displayed on a display device in real time or near real time (e.g., the image is displayed as soon as it is generated and there is no intentional delay). The operator of the ultrasound probe can view the image and adjust various acquisition parameters and / or positions of the ultrasound probe in order to obtain a high-quality image of the target anatomical feature structure (e.g., heart, liver, kidney, or another anatomical feature structure). Adjustable acquisition parameters include transmission frequency, transmission depth, gain (e.g., total gain and / or time gain compensation), cross beam, beam steering angle, beam forming strategy, frame averaging, and / or other parameters. Changing acquisition parameters to obtain the best image (e.g., an image with a desired quality) may be very challenging and based on user experience. In addition, it may be difficult to separate image quality problems caused by probe placement from image quality problems caused by acquisition parameters. Therefore, the operator's adjustment of acquisition parameters to obtain the best image is usually subjective. For example, the operator may adjust various acquisition parameters and / or adjust the placement of the probe until an image is acquired that appears optimal to the operator, and the process may not be defined or repeated between exams. This subjectivity and lack of a deterministic process may lead to non-reproducible results, and in many ultrasound exams, the highest quality image possible may not be acquired.

[0014] Thus, according to embodiments disclosed herein, a set of image quality models may be used to assess various aspects of image quality, the set of image quality models being configured to distinguish probe placement errors from improper acquisition parameters and provide guidance to an operator of an ultrasound system in order to generate the best possible image of an imaged anatomical structure. The set of image quality models may include an anatomical model that is trained to determine whether all expected anatomical features are present in an image and may be deployed to identify whether an ultrasound probe is correctly positioned to an anatomical structure specified by an image. The set of image quality models may also include one or more scan setting models, each of which is trained to determine whether an image has been acquired with appropriate / sufficient scan settings. For example, one or more scan setting models may determine whether transmit frequency, transmit depth, gain, beam steering angle, beamforming strategy, cross beam, frame averaging, and / or other settings are appropriate / sufficient for one or more diagnostic objectives (e.g., measurement of indicated anatomical features, visualization of indicated anatomical features, and / or automated tasks such as segmentation, determination of findings, etc.). If the set of image quality models identifies a probe placement or scan setup problem, guidance may be output to an operator of the ultrasound system such that the probe may be repositioned and / or one or more acquisition parameters may be adjusted. Identification of issues causing poor image quality and guidance on how to resolve image quality issues as described herein may simplify operator workflow, which may reduce exam time and may facilitate higher quality exams, even for more inexperienced operators. Furthermore, identification of issues causing poor image quality and guidance on how to resolve image quality issues as described herein may increase exam consistency between patients, and even between scans of the same patient.

[0015] Figure 1 An exemplary ultrasound system is shown in FIG. , which includes an ultrasound probe, a display device, and an imaging processing system. Via the ultrasound probe, an ultrasound image can be acquired and displayed on the display device. As described above, various acquisition scanning parameters (such as frequency and depth) can be used to acquire the image. In order to determine whether the acquired ultrasound image has sufficient image quality, the ultrasound image can be input into a set of cascaded image quality models, which can determine whether the unsatisfactory (e.g., low) image quality is due to probe placement, acquisition parameter settings, or both. In addition, the image quality model can identify which acquisition parameters cause the unsatisfactory image quality. As shown in FIG. Figure 2 As shown, the image processing system includes image quality assessment models (such as an anatomical model and one or more scan setting models), which can be deployed according to Figure 4 and Figure 5 The method shown evaluates Figure 3The image quality problems shown are caused by poor probe placement and / or acquisition parameters. Figure 6 In some examples shown, the scan settings model can utilize an uncertainty network. Output from the image quality model can be used to generate guidance / notifications that can be output for display to a user, such as Figures 7 to 10 An exemplary graphical user interface is shown.

[0016] See also Figure 1 , a schematic diagram of an ultrasound imaging system 100 according to an embodiment of the present disclosure is shown. The ultrasound imaging system 100 includes a transmit beamformer 101 and a transmitter 102, which drives an element (e.g., a transducer element) 104 within a transducer array (referred to herein as a probe 106) to transmit a pulsed ultrasound signal (referred to herein as a transmit pulse) into a body (not shown). According to one embodiment, the probe 106 can be a one-dimensional transducer array probe. However, in some embodiments, the probe 106 can be a two-dimensional matrix transducer array probe. As further explained below, the transducer element 104 can be made of a piezoelectric material. When a voltage is applied to a piezoelectric crystal, the crystal physically expands and contracts, thereby emitting an ultrasonic ball wave. In this way, the transducer element 104 can convert an electronic transmit signal into an acoustic transmit beam.

[0017] After the elements 104 of the probe 106 transmit the pulsed ultrasound signals into the body (of the patient), the pulsed ultrasound signals are scattered back from structures inside the body (such as blood cells or muscle tissue) to produce echoes that return to the elements 104. The echoes are converted into electrical signals or ultrasound data by the elements 104, and the electrical signals are received by the receiver 108. The electrical signals representing the received echoes pass through the receive beamformer 110 which outputs the ultrasound data. In addition, the transducer elements 104 may generate one or more ultrasound pulses based on the received echoes to form one or more transmit beams.

[0018] According to some embodiments, the probe 106 may include electronic circuits to perform all or part of the transmit beamforming and / or receive beamforming. For example, all or part of the transmit beamformer 101, the transmitter 102, the receiver 108, and the receive beamformer 110 may be located within the probe 106. In the present disclosure, the term "scan" or "scanning" may also be used to refer to the acquisition of data by the process of transmitting and receiving ultrasound signals. In the present disclosure, the term "data" may be used to refer to one or more data sets acquired by the ultrasound imaging system. In one embodiment, the data acquired via the ultrasound system 100 may be used to train a machine learning model. The user interface 115 may be used to control the operation of the ultrasound imaging system 100, including input for controlling patient data (e.g., patient medical history), for changing scanning or display parameters, for starting a probe repolarization sequence, etc. The user interface 115 may include one or more of the following: a rotating element, a mouse, a keyboard, a trackball, a hard key linked to a specific action, a soft key that can be configured to control different functions, and / or a graphical user interface displayed on a display device 118.

[0019] The ultrasound imaging system 100 also includes a processor 116, which is used to control the transmit beamformer 101, the transmitter 102, the receiver 108, and the receive beamformer 110. The processor 116 is in electronic communication (e.g., communicatively connected) with the probe 106. For the purpose of this disclosure, the term "electronic communication" may be defined as including both wired communication and wireless communication. The processor 116 can control the probe 106 to collect data according to instructions stored in the processor's memory and / or the memory 120. The processor 116 controls which of the elements 104 are active and the shape of the beam emitted from the probe 106. The processor 116 also communicates electronically with the display device 118, and the processor 116 can process data (e.g., ultrasound data) into an image for display on the display device 118. The processor 116 may include a central processing unit (CPU) according to one embodiment. According to other embodiments, the processor 116 may include other electronic components capable of performing processing functions, such as a digital signal processor, a field programmable gate array (FPGA), or a graphics board. According to other embodiments, the processor 116 may include a plurality of electronic components capable of performing processing functions. For example, the processor 116 may include two or more electronic components selected from a list of electronic components, including: a central processing unit, a digital signal processor, a field programmable gate array, and a graphics board. According to another embodiment, the processor 116 may also include a composite demodulator (not shown) that demodulates the RF data and generates raw data. In another embodiment, the demodulation may be performed earlier in the processing chain. The processor 116 is adapted to perform one or more processing operations according to a plurality of optional ultrasound modalities on the data. In one example, the data may be processed in real time during a scanning session because the echo signal is received by the receiver 108 and transmitted to the processor 116. For the purposes of this disclosure, the term "real time" is defined as including a process performed without any intentional delay. For example, an embodiment may acquire images at a real-time rate of 7 to 20 frames / second. The ultrasound imaging system 100 is capable of acquiring 2D data of one or more planes at a significantly faster rate. However, it should be understood that the real-time frame rate may depend on the length of time it takes to acquire each frame of data for display. Therefore, when a relatively large amount of data is acquired, the real-time frame rate may be slower. Thus, some embodiments may have real-time frame rates significantly faster than 20 frames / second, while other embodiments may have real-time frame rates less than 7 frames / second. Data may be temporarily stored in a buffer (not shown) during a scanning session and processed in a less real-time manner in real-time or off-line operation. Some embodiments of the present invention may include multiple processors (not shown) to handle processing tasks handled by processor 116 according to the exemplary embodiments described above.For example, a first processor may be utilized to demodulate and extract the RF signal prior to displaying the image, while a second processor may be utilized to further process the data (eg, by augmenting the data as further described herein). It will be appreciated that other embodiments may utilize different processor arrangements.

[0020] The ultrasound imaging system 100 can continuously acquire data at a frame rate of, for example, 10 Hz to 30 Hz (e.g., 10 to 30 frames per second). The image generated based on the data can be refreshed at a similar frame rate on the display device 118. Other embodiments can acquire and display data at different rates. For example, depending on the size of the frame and the intended application, some embodiments can acquire data at a frame rate less than 10 Hz or greater than 30 Hz. A memory 120 is included for storing frames of processed acquisition data. In an exemplary embodiment, the memory 120 has sufficient capacity to store at least a few seconds of ultrasound data frames. The data frames are stored in a manner that facilitates retrieval according to their acquisition order or time. The memory 120 may include any known data storage medium.

[0021] In various embodiments of the present invention, the processor 116 can process data through different mode-related modules (e.g., B-mode, color Doppler, M-mode, color M-mode, spectral Doppler, elastic imaging, TVI, strain, strain rate, etc.) to form 2D or 3D data. For example, one or more modules can generate B-mode, color Doppler, M-mode, color M-mode, spectral Doppler, elastic imaging, TVI, strain, strain rate, and combinations thereof, etc. As an example, one or more modules can process color Doppler data, which may include traditional color blood flow Doppler, power Doppler, HD flow, etc. The image lines and / or frames are stored in the memory and may include timing information indicating the time when the image lines and / or frames are stored in the memory. These modules may include, for example, a scan conversion module for performing a scan conversion operation to convert the acquired image from beam space coordinates to display space coordinates. A video processor module may be provided that reads the acquired image from the memory and displays the image in real time while a procedure (e.g., ultrasound imaging) is performed on the patient. The video processor module may include a separate image memory, and the ultrasound image may be written to the image memory for reading and display by the display device 118 .

[0022] In various embodiments of the present disclosure, one or more components of the ultrasound imaging system 100 may be included in a portable handheld ultrasound imaging device. For example, the display device 118 and the user interface 115 may be integrated into the external surface of the handheld ultrasound imaging device, which may also include a processor 116 and a memory 120. The probe 106 may include a handheld probe that electronically communicates with the handheld ultrasound imaging device to collect raw ultrasound data. The transmit beamformer 101, the transmitter 102, the receiver 108, and the receive beamformer 110 may be included in the same or different parts of the ultrasound imaging system 100. For example, the transmit beamformer 101, the transmitter 102, the receiver 108, and the receive beamformer 110 may be included in a handheld ultrasound imaging device, a probe, and a combination thereof.

[0023] After performing a two-dimensional ultrasound scan, a data block containing the scan lines and their samples is generated. After applying the back-end filter, a process called scan conversion is performed to transform the two-dimensional data block into a displayable bitmap image with additional scanning information (such as depth, the angle of each scan line, etc.). During scan conversion, interpolation techniques are applied to fill in the missing holes (i.e., pixels) in the resulting image. These missing pixels occur because each element of the two-dimensional block should generally cover many pixels in the resulting image. For example, in current ultrasound imaging systems, bicubic interpolation is applied, which utilizes adjacent elements of the two-dimensional block. Therefore, if the two-dimensional block is relatively small compared to the size of the bitmap image, the scan-converted image will include areas of poor or low resolution, especially for areas with greater depth.

[0024] The ultrasound images acquired by the ultrasound imaging system 100 may be further processed. In some embodiments, the ultrasound images generated by the ultrasound imaging system 100 may be transmitted to an image processing system, where in some embodiments, the ultrasound images may be analyzed by one or more machine learning models trained using the ultrasound images and corresponding ground truth outputs to assess image quality issues of the ultrasound images. As used herein, the ground truth output refers to the expected or "correct" output based on a given input into the machine learning model. For example, if a machine learning model is being trained to classify images of cats, then when an image of a cat is fed, the ground truth output of the model is the label "cat". As explained in more detail below, if a machine learning model is being trained to classify ultrasound images based on adequate imaging of target anatomical features, the ground truth output of the model may be a label indicating the presence and extent of the target anatomical features.

[0025] Although described herein as separate systems, it should be understood that in some embodiments, the ultrasound imaging system 100 includes an image processing system. In other embodiments, the ultrasound imaging system 100 and the image processing system may include separate devices. In some embodiments, the images generated by the ultrasound imaging system 100 may be used as a training data set for training one or more machine learning models, where the machine learning model may be used to perform one or more steps of ultrasound image processing as described below.

[0026] See also Figure 2 , which shows an image processing system 202 according to an embodiment. In some embodiments, the image processing system 202 is incorporated into the ultrasound imaging system 100. For example, the image processing system 202 may be provided in the ultrasound imaging system 100 as a processor 116 and a memory 120. In some embodiments, at least a portion of the image processing system 202 is provided at a device (e.g., an edge device, a server, etc.) that is communicatively coupled to the ultrasound imaging system via a wired connection and / or a wireless connection. In some embodiments, at least a portion of the image processing system 202 is provided at a separate device (e.g., a workstation) that can receive images from the ultrasound imaging system or from a storage device storing images / data generated by the ultrasound imaging system. The image processing system 202 may be operably / communicatively coupled to a user input device 232 and a display device 234. At least in some examples, the user input device 232 may include a user interface 115 of the ultrasound imaging system 100, and the display device 234 may include a display device 118 of the ultrasound imaging system 100.

[0027] Image processing system 202 includes processor 204, and this processor is configured to execute the machine-readable instruction stored in non-transitory memory 206.Processor 204 can be single-core or multi-core, and the program executed thereon can be configured to carry out parallel or distributed processing.In some embodiments, processor 204 can optionally include separate components distributed in two or more devices, and these separate components can be remotely located and / or be configured for coordinated processing.In some embodiments, one or more aspects of processor 204 can be virtualized and performed by a remotely accessible networked computing device configured with a cloud computing configuration.

[0028] The non-transitory memory 206 may store an image quality model 208, a training module 210, and ultrasound image data 212. The image quality module 208 may include one or more machine learning models (such as deep learning networks) including a plurality of weights and biases, an activation function, a loss function, a gradient descent algorithm, and instructions for implementing the one or more deep neural networks to process an input ultrasound image. For example, the image quality model 208 may store instructions for implementing an anatomical model 209 and / or one or more scan setting models 211. The anatomical model 209 and the one or more scan setting models 211 may each include one or more neural networks. The image quality module 208 may include trained and / or untrained neural networks and may also include training routines or parameters (e.g., weights and biases) associated with one or more neural network models stored therein.

[0029] The anatomical model 209 may be a neural network (e.g., a convolutional neural network) that is trained to recognize whether an expected anatomical feature is present in an ultrasound image and whether the expected anatomical feature has expected morphological characteristics. The anatomical model 209 may be trained to recognize expected anatomical features in a scan plane / view specific manner. For example, the anatomical model may be trained to recognize expected anatomical features in a four-chamber view of the heart, but not in a parasternal long axis (PLAX) view of the heart. Thus, in some examples, the anatomical model 209 may actually include multiple anatomical models, each of which is specific to a different scan plane or anatomical view. The anatomical model 209 may be trained to output a probe position quality metric that reflects the quality of an input ultrasound image that varies with anatomical adequacy (e.g., whether the expected anatomical feature is present and has the expected morphological characteristics for the scan plane currently being imaged). For example, expected anatomical features identified by the anatomical model may change in appearance, position, and / or presence as the probe position changes, and the probe position quality metric output by the anatomical model 209 may reflect the appearance / visibility of these anatomical features as an indicator of whether the probe is in an optimal / correct position for acquiring ultrasound images. As used herein, a probe being in an optimal / correct position (or non-optimal or incorrect position) may refer to positioning the probe based on a target (e.g., standard) ultrasound scan plane so that the expected anatomical features are not present in the acquired image. Thus, the determination of whether the probe is placed in a good or poor position is based on whether the expected or expected anatomical features are present in the image of a given target scan plane.

[0030] One or more scan setting models 211 (also referred to herein as acquisition setting models) may include one or more neural networks or other machine learning models that are trained to output corresponding acquisition setting quality metrics that represent image quality factors that vary with acquisition parameter values. One or more scan setting models 211 may include a first scan setting model (referred to as a depth model) that evaluates depth, a second scan setting model (referred to as a gain model) that evaluates gain, a third scan setting model (referred to as a frequency model) that evaluates frequency, and / or additional scan setting models that evaluate other acquisition parameters (e.g., time gain compensation (TGC), frame averaging, etc.). The depth model may be trained to output a depth image quality metric that reflects the level of impact of the depth setting on the image quality of the ultrasound image. The gain model may be trained to output a gain image quality metric that reflects the level of impact of the gain setting on the image quality of the ultrasound image. The frequency model may be trained to output a frequency image quality metric that reflects the level of impact of the frequency setting on the image quality of the ultrasound image. Each of the depth model, gain model, frequency model, and / or any additional scan setting models may include a respective uncertainty network that is trained to perform a proxy task on an input ultrasound image, such as segmenting a particular anatomical feature. The amount of uncertainty in segmenting the anatomical feature may be assessed and associated with particular acquisition parameters and quantified to generate a second image quality metric, as described in more detail below.

[0031] The non-transitory memory 206 may also include a training module 210 including instructions for training one or more machine learning models stored in the image quality model 208. In some embodiments, the training module 210 is not located at the image processing system 202. Thus, the image quality model 208 includes a trained and validated network.

[0032] The non-transitory memory 206 may also store ultrasound image data 212, such as Figure 1The ultrasound image data 212 may include ultrasound image data acquired by the ultrasound imaging system 100. For example, the ultrasound image data 212 may include ultrasound image data acquired by the ultrasound imaging system 100. The ultrasound images of the ultrasound image data 212 may include ultrasound images acquired by the ultrasound imaging system 100 under different scanning settings, such as different frequencies, different gains, different depths, different TGCs, different frame averages, etc. In addition, when the training module 210 is stored in the non-transitory memory 206, the ultrasound image data 212 may store ultrasound images, ground truth outputs, iterations of machine learning model outputs, and other types of ultrasound image data that can be used to train the image quality model 208. In some embodiments, the ultrasound image data 212 may store ultrasound images and ground truth outputs in an ordered format, such that each ultrasound image is associated with one or more corresponding ground truth outputs. However, in an example where the training module 210 is not provided at the image processing system 202, images / ground truth outputs that can be used to train the image quality model 210 may be stored elsewhere.

[0033] In some embodiments, non-transitory memory 206 may include components disposed on two or more devices that may be remotely located and / or configured for coordinated processing. In some embodiments, one or more aspects of non-transitory memory 206 may include a remotely accessible networked storage device configured in a cloud computing configuration.

[0034] The user input device 232 may include one or more of a touch screen, keyboard, mouse, trackpad, motion sensing camera, or other device configured to enable a user to interact with and manipulate data within the image processing system 202. In one example, the user input device 232 may enable a user to select an ultrasound image for use in training a machine learning model to indicate or mark the location of an interventional device in the ultrasound image data 212, or for further processing using a trained machine learning model.

[0035] The display device 234 may include one or more display devices utilizing almost any type of technology. In some embodiments, the display device 234 may include a computer monitor and may display ultrasound images. The display device 234 may be combined with the processor 204, the non-transitory memory 206, and / or the user input device 232 in a shared housing, or may be a peripheral display device and may include a monitor, a touch screen, a projector, or other display devices known in the art that may enable a user to view ultrasound images generated by the ultrasound imaging system and / or interact with various data stored in the non-transitory memory 206.

[0036] It should be understood that Figure 2The image processing system 202 shown is for illustration and not limitation. Another suitable image processing system may include more, fewer, or different components.

[0037] Go to Figure 3 , which shows multiple ultrasound images 300 that depict different image quality issues in the probe placement domain and the scan setting domain of a fetal central nervous system (CNS) scan. Image 302 is a standard scan image, i.e., the result of both good probe placement and good scan settings. In image 302, two anatomical structures of the brain are highlighted: the cavum septum pellucidum (large rectangle) and the gyri (small rectangle). The rest of the figure shows examples of lower quality images separated by domain and represented on two axes, where a first group of images 304 on the vertical axis represent examples of images with insufficient / low image quality due to improper probe placement, and a second group of images 312 on the horizontal axis represent examples of images with low image quality due to improper scan settings. As used herein, low or poor image quality due to improper or inadequate scan settings may refer to the scan settings (e.g., acquisition parameters such as depth, frequency, gain, etc.) of the ultrasound probe used to acquire the image such that desired anatomical features are not sufficiently visible in the image and / or such that desired anatomical features do not have sufficiently well-defined boundaries in the image and / or image contrast, image brightness, image saturation, etc. are not at desired levels and / or are not consistent throughout the image. Thus, a determination of whether an image has low quality due to scan settings may be based on whether an expert (e.g., a trained clinician) would indicate that the image is suitable for a given diagnostic purpose (e.g., diagnosing or ruling out a condition, lesion detection, etc.).

[0038] In the probe placement domain (vertical axis), factors affecting scan plane quality include structural morphological changes and anatomical structure loss. Image 310 and image 308 show anatomical structure loss; in image 310, the cavum septum pellucidum is present but the gyrus is not present, while in image 308, the gyrus is present but the cavum septum pellucidum is not present. Image 306 shows structural changes. For example, although both target anatomical features (e.g., cavum septum pellucidum and gyrus) are present in image 308, each anatomical feature has a different morphology than the corresponding anatomical feature in image 302.

[0039] On the other hand, suboptimal scan settings can cause a perceived loss of information. Suboptimal time gain compensation (TGC) or gain settings can blur structures in the field of view, while suboptimal depth settings can exclude structures from the field of view. Examples of image quality issues in the scan settings domain (horizontal axis) caused by inadequate gain and suboptimal depth / TGC settings are shown in the second set of images 312. Images 314, 316, and 318 show examples of suboptimal depth, inadequate gain, and incorrect TGC, respectively.

[0040] In addition, although not Figure 3 , but image quality may be affected by both probe position and scan settings. Thus, expected anatomical features may be missing and / or exhibit morphological changes due to poor probe placement, while there may be a perceived loss of information due to suboptimal scan settings. In some examples, standard image quality assessment may have difficulty distinguishing image quality issues due to probe placement from image quality issues due to scan settings, making it challenging for an inexperienced ultrasound operator to adjust probe placement and / or scan settings when faced with low-quality images. Therefore, as explained in more detail below, a series of image quality assessment models may be deployed to assess image quality, distinguish poor probe placement from suboptimal scan settings, and / or provide guidance to operators of ultrasound systems to obtain high-quality images in a reproducible manner.

[0041] Figure 4 A flow chart illustrating an exemplary method 400 for assessing image quality of an ultrasound image is shown, according to one embodiment. Figure 1 to Figure 2 The method 400 is described with reference to a system and components of FIG. 1 , but it should be understood that the method 400 may be implemented with other systems and components without departing from the scope of the present disclosure. The method 400 may be implemented based on a computer stored in a computing device such as Figure 2 The instructions in the non-volatile memory of the image processing system 202 are executed.

[0042] At 402, an ultrasound image is acquired and displayed on a display device. Figure 1The ultrasound probe 106 acquires an ultrasound image and displays it to the operator via the display device 118. The image may be acquired and displayed in real time or near real time, and may be acquired by default or user-specified scanning parameters (e.g., default depth, frequency, etc.). At 404, the method 400 determines whether a request to evaluate the image quality of the ultrasound image has been received. The request may be received automatically (e.g., as part of a scanning protocol) or may be received via user input from an operator of the ultrasound system. The request may include an indication of a target scanning plane for the ultrasound image. For example, an ultrasound image may be acquired as part of an ultrasound examination, in which certain anatomical features are imaged in certain views / axes in order to diagnose a patient's condition, measure various aspects of anatomical features, etc. For example, during a cardiac examination, one or more target scanning planes (also referred to as views) of the patient's heart may be imaged. The target scanning planes may include a four-chamber view, a two-chamber view (which may also be referred to as a short-axis view), and a long-axis view (which may also be referred to as a PLAX view or a three-chamber view). One or more images may be acquired in each scanning plane and saved as part of the examination for later analysis by a clinician (such as a cardiologist). When acquiring images for examination, an ultrasound operator (e.g., a sonographer) can move the ultrasound probe until the operator determines that a target scan plane is being imaged, and then the operator can input an input indicating that the target scan plane is being imaged and requesting an evaluation of the image quality of the acquired ultrasound image of the target scan plane (e.g., via the user interface 115). In another example, the ultrasound system (e.g., via the image processing system 202) can automatically determine that the target scan plane is being imaged. For example, each acquired ultrasound image (or a certain frequency of acquired ultrasound images, such as every five images) can be input into a detection model configured to automatically detect the current scan plane. If the current scan plane is the target scan plane, an indication can be generated.

[0043] If a request to evaluate image quality is not received, the method 400 returns to 402 to continue acquiring and displaying ultrasound images (e.g., with default or user-set scanning parameters). If a request to evaluate image quality is received, the method 400 proceeds to 406 to input the ultrasound image as input to an anatomical model (such as a Figure 2 The anatomical model 209 may be trained to recognize individual anatomical structures, thereby outputting an indication of whether a given image has insufficient anatomical structures due to the absence of any key / expected anatomical structures and / or morphological changes in key / expected anatomical structures. The anatomical model may include multiple components, each of which is trained to recognize a specific anatomical structure. For example, Figure 3As seen in image 302 of , a fetal CNS anatomical model that takes as input an ultrasound image of the brain may include one component (e.g., a network) trained to identify the cavum septum pellucidum and another component (e.g., a network) trained to identify the gyri (or one component trained to identify both the cavum septum pellucidum and the gyri). The anatomical model may output an indication of whether the input image includes sufficient anatomical features (based on knowledge of the target scan plane and therefore what anatomical features are expected to be present), such as a binary yes / no value indicating sufficient anatomical structures, a list of anatomical structures present in the image that can be compared to a list of expected features, a visual indication of anatomical structures that can be identified, or another similar indication of scan plane quality. In addition, a component (e.g., a network) of the anatomical model may output an indication of any change in structural morphology, either as a numerical value indicating a degree of distortion, or as a binary yes / no result indicating whether the structural change falls within an acceptable range of scan plane quality. This output may be displayed on a display device as a notification of probe misplacement and may include guidance with respect to moving the probe, an example of which is seen in Fig.10 middle.

[0044] At 408, method 400 evaluates whether the model output indicates that sufficient anatomy is present in the input image. When all expected anatomy is visible in the scan plane and any visible structural changes fall within an acceptable range, the anatomy can be determined to be sufficient. If the model output does not indicate sufficient anatomy, method 400 proceeds to 410 to output a notification and / or instructions to move the probe, and method 400 returns. The notification and / or instructions to move the probe can be output for display on a display device. The notification may include, for example, Fig.10 A textual indication of a specific quality issue shown in notification 1004 in (e.g., an indication that an anatomical structure is missing), one or more display elements indicating sufficient anatomical structure in a binary or non-binary manner, one or more display elements indicating structural changes of anatomical features in a binary or non-binary manner, or a simple indication of poor scan plane quality or a need for probe adjustment. In this way, probe position quality parameters of an input ultrasound image can be determined via the anatomical model, where the probe position quality parameters represent ultrasound probe positioning errors. If the probe position quality parameters (e.g., output from the anatomical model) indicate that the ultrasound probe is correctly positioned, the operator can be informed of the correct positioning and / or that the expected anatomical features are visible in the input image. If the probe position quality parameters indicate poor probe positioning (e.g., because the expected anatomical features are not visible and / or the expected anatomical features have an unexpected morphological appearance), a notification can be output for display.

[0045] If the output from the anatomical model indicates that sufficient anatomy is present, the method 400 proceeds to 412 to input the image as input to a scan setup model (such as Figure 2 The scanning setting model 211). Figure 5Additional details about the scan settings model are presented. Briefly, the scan settings model may generate an output based on an input image, wherein the output indicates a quality level of the image for each of one or more scan settings. For example, the scan settings model may include a first scan settings model that generates an output indicating an image quality level due to gain, a second scan settings model that generates an output indicating an image quality level due to depth, and the like. Thus, the scan settings model may output one or more acquisition settings quality parameters of the input ultrasound image, wherein each acquisition settings quality parameter represents a corresponding quality level of the ultrasound image relative to a corresponding acquisition setting used to acquire the ultrasound image, wherein the acquisition parameters may include depth, frequency, gain, and the like.

[0046] At 414, method 400 determines whether the model output from the scan settings model indicates adequate scan settings. The scan settings model can be trained, at least in some examples, to define individual anatomical structures by identifying their contours (segmentation), and the scan settings model can output an indication of whether a given image has insufficient / low image quality due to suboptimal scan settings (e.g., oversaturation or low gain in the far field). The scan settings model can include multiple components, each of which is trained to define a particular anatomical structure. For example, Figure 3 As seen in image 302 of , a fetal CNS scan setup model that takes an ultrasound image of the brain as input may include one component model trained to segment the cavum septum pellucidum and another component trained to segment the gyrus, or one component model trained to segment both. Each component model may output a numerical value indicating the degree of uncertainty of the model in segmenting the corresponding anatomical structure, as measured by the standard deviation within n segmentation attempts, wherein a higher standard deviation indicates that the model has difficulty clearly segmenting the corresponding anatomical structure, whereby a higher uncertainty value represents an inadequate scan setup relative to the corresponding anatomical structure.

[0047] The anatomical model may output a numerical value indicative of the overall uncertainty of the model in performing a segmentation task within a set of anatomical structures as a function of the individual uncertainty values ​​produced by its different components. This numerical value may be output for display on a display device as an indication of scan plane quality. In other embodiments, this value may be compared to a scan plane quality threshold and used to output a binary yes / no indication of whether the scan plane is of sufficient quality or whether scan settings may be adjusted.

[0048] If it is determined that the scan settings are insufficient, the method 400 proceeds to 416 to output a notification and / or instructions for adjusting the settings, and the method 400 returns. In one embodiment, the notification output at 416 informs the operator whether the poor image quality is such that Figures 7 to 10The results of suboptimal depth settings, suboptimal TGC settings, inadequate gain, and / or other scan setting issues (e.g., frequency) are shown. If the model output indicates at 414 that the scan settings are adequate, the method 400 proceeds to 418 to output a notification indicating the adequacy of both the anatomy and the scan settings, and then returns. Figure 6 An example of a notification indicating sufficient anatomy and scan settings is shown in and described below. In this way, feedback can be output to a user of the ultrasound system based on a probe position quality parameter (e.g., output from an anatomical model) and / or one or more acquisition setting quality parameters (e.g., output from a scan settings model). In addition, in some examples, if an input image is determined to have sufficient anatomy and sufficient scan settings, an additional ultrasound image in a scan plane with the current scan settings can be acquired, the image (and / or any subsequently acquired image) can be input into a subsequent step of the scan protocol (e.g., automatic measurements can be performed on the image), and / or the image can be saved as part of an examination for subsequent clinician review.

[0049] Now go to Figure 5 , presents a method 500 for determining an image quality metric using a scan settings model. The method 500 may be performed as part of the method 400, for example, the method 500 may be performed to generate a model output evaluated at 414. The method 500 may be performed based on a computer program stored on a computing device such as Figure 2 The method 500 is executed by instructions in a non-transitory memory of the image processing system 202 of the present invention. As presented herein, the method 500 is specific to determining image quality caused by ultrasound scan settings of one scan parameter (e.g., TGC). It should be understood that the process set forth in the method 500 can be applied to additional scan parameters (e.g., depth, frequency, etc.), as will be explained in more detail below. Therefore, the method 500 can be performed for each different scan setting model described herein.

[0050] Method 500 may identify whether images of a given scan plane were acquired with inadequate / suboptimal scan settings. Prior to deploying the scan settings model, the scan settings model may be trained on a proxy task using sample images of a given scan plane acquired at optimal scan settings as input, as indicated at 502. The scan settings model may include multiple component models, each trained on a different task and / or different scan settings, whereby the scan settings model outputs results that vary with the outputs produced by its component models. In training a scan settings network for gain, for example, the training data consists of images with optimal gain, but the scan settings for other parameters (frequency, depth, etc.) are not necessarily optimal in the images of the training data. Similarly, for training a deep model, the deep model training data consists of images with optimal depth but not necessarily optimal frequency, gain, TGC, etc., so that each model may be trained using only images with the optimal settings for that model while providing variability in other acquisition parameters.

[0051] Anatomical segmentation is used as an exemplary proxy task herein, whereby a component model is trained to segment a specific anatomical structure in a given scan plane. In other embodiments, the proxy task may involve assessing other characteristics of the anatomical structure (such as position or relative size) or characteristics of the image itself (such as orientation, clarity or distortion of features or regions), etc. In general, the proxy task may involve any operation that can be performed on an ultrasound image for which training data is available (e.g., anatomical scan plane classification, structure localization, etc.).

[0052] It should be understood that 502 is optionally included within method 500, as indicated by the dashed line. The model training described at 502 may instead be performed on a different device prior to performing method 500. For example, the scan settings model may be trained on a remote device (e.g., server, cloud), and once the scan settings model is validated, the scan settings model may be deployed on the image processing system 202. In a further example, the scan settings model may reside on a remote device, and the scan settings model may receive an input image from the image processing system 202 and send a generated output to the image processing system 202.

[0053] At 504, an image is received as input. At 506, a proxy task (e.g., segmentation) is performed on the image n times using a variational network to produce n plausible results (segments). At 508, the uncertainty of the n results of the proxy task is determined. For example, the ability of the model to consistently perform the proxy task within the n results is quantified via calculation of a standard deviation based on the average segmentation, thereby outputting a numerical value or other suitable value representing the degree of uncertainty that the model is able to perform the proxy task ( Figure 6The uncertainty determination is used to generate a model output at 510, where relatively high uncertainty indicates that the scan settings may be adjusted. In one embodiment, the model output is displayed via a series of separate visual elements that distinguish between uncertainties caused by incorrect or insufficient TGC, gain, or depth, and the degree of uncertainty is shown via a color gradient, such as Figures 7 to 10 In other embodiments, uncertainty may be expressed in a binary fashion as a given threshold being exceeded for each individual factor or as a collective quantification of the overall uncertainty.

[0054] exist Figure 6 In , schematic diagram 600 shows exemplary inputs and outputs of a variational network included as part of an embodiment of a scan setting model, which is herein an uncertainty network as part of a gain model for evaluating whether the current gain setting contributes to low or high image quality. A sample input image 602 is fed into a proxy task uncertainty network 604 to perform segmentation of the anatomical structure of the septum in a four-chamber (4ch) view / scan plane of the heart. Image 612 shows the septum in a 4ch view under optimal settings (e.g., the scan settings required to produce the target scan plane). Multiple results 606 show the output of the segmentation task, including n plausible segmentations that vary to some extent. The multiple results are the result of the configuration of the variational neural network, which produces multiple outputs as the results of the probabilistic interpretation. From these outputs, the average segmentation value (μ) 608 and the standard deviation (σ) 610 are obtained. The standard deviation is a measure of the inconsistency between different segmentation outputs (in this case relative to the TGC setting). The standard deviation is used as a quantification of uncertainty, whereby low uncertainty (standard deviation) indicates images with TGC settings similar to those in the training cohort.

[0055] Figures 7 to 10 Examples of graphical user interfaces (GUIs) are shown that can be output as Figure 4 800. GUI 700 includes an image 702 acquired with optimal scan settings and probe placement. A set of display elements 704 shows four separate visual elements corresponding to depth, frequency, TGC, and gain, the parameters of which have been generated by method 400 (including method 500 performed as part of method 400) based on the determination of uncertainty as described above. In 704, low uncertainty values ​​are associated with all four parameters, so no visual indication of uncertainty is displayed. In contrast, in GUI 800, image 802 is the result of a suboptimal TGC setting communicated to the operator via visual element 804 (also in FIG. 804). Figure 3318), thereby displaying a grayscale gradient corresponding to the TGC uncertainty expressed as a percentage. Other embodiments may use color or other visual means to indicate the degree of uncertainty generated by method 400 relative to a standard scan with optimal scan settings and probe placement.

[0056] Likewise, GUI 900 includes Figure 3 902, which corresponds to image 314 in FIG. 8, has a low quality due to a suboptimal depth setting. As with image 804, an inadequate depth setting may be communicated via visual element 904, which indicates a high uncertainty value of the depth expressed as a percentage in a grayscale gradient. GUI 1000 includes Figure 3 308 in , the image 1002 corresponding to the image 308 in , which shows the change in structural morphology caused by poor probe placement (no transparent cavity). In this example, the notification is displayed as a visual element 1004. In the example shown, the visual element includes a notification that the expected anatomical feature (transparent cavity) is not present, but other notifications are also possible. For example, the operator may be notified of the insufficient anatomical structure via other display elements of a binary or non-binary nature, such as a check box, an indication of the anatomical structure present in the image compared to the expected feature, a visual indication of the anatomical structure that can be identified (such as a rectangle drawn in image 1002), an indication of any change in structural morphology in binary or non-binary form, or any other alternative display element that effectively conveys the insufficient anatomical structure. In addition, in some examples, guidance may be provided to the operator, such as an instruction to move the probe.

[0057] Therefore, the systems and methods described herein provide for evaluating the image quality of one or more ultrasound images in a manner that allows probe placement issues to be distinguished from scan setup issues. Currently, image quality metrics are typically calculated independently of the clinician's task / imaging diagnostic goals and driven by physics / image processing (such as contrast-to-noise ratio, etc.). However, the clinician's ideal image is not always an image with the best metric values ​​(SNR, CNR, etc.). For example, in order to evaluate the gallbladder, the clinician needs to observe the bright / saturated tail along the sound path. In addition, current image quality indices are typically scalars. For scan planes that are determined to be poor, the quality index does not explain the root cause of the low image quality, such as whether it is probe placement related or scan setup related. Therefore, the user may not be able to decide on remedial actions to improve the quality of the scan plane.

[0058] The systems and methods described herein distinguish probe placement-related image degradation from scan setting-related image degradation. Inappropriate probe placement is identified herein due to changes in structures seen in the scan plane (loss of structure or morphological changes). Suboptimal scan settings are identified herein based on the perceived loss of information. For example, suboptimal TGC / gain settings can obscure structures in the field of view, incorrect depth can exclude structures from the field of view, etc.

[0059] As described above, uncertainty networks included as part of one or more image quality models may be used herein to identify areas where scan settings are suboptimal in a given scan plane. To achieve this, the network is trained to perform a proxy task on a given scan plane using only sample images produced by optimal scan settings for the target parameters (although low-quality sample images caused by suboptimal scan settings for other parameters are included). These models quantify the uncertainty of the network in performing the proxy task when deployed. Areas of high uncertainty indicate a potential need for changes in scan settings, as higher uncertainty corresponds to a need to change settings.

[0060] Clinical diagnosis of biometric aspects of ultrasound images requires the acquisition of correct / standard scan planes. Standard views provide ease of assessment due to the presentation of key anatomical structures. Consistently reaching standard scan planes requires high skill on the part of the ultrasound practitioner. Even if an anatomically correct scan plane has been reached, incorrect scan settings can obscure key structures (oversaturation, low gain, incorrect TGC, insufficient depth, etc.).

[0061] Acquisition of diagnostic scan planes can be affected by poor probe placement and / or suboptimal scanner settings, resulting in poor visibility of critical anatomical structures. Conventional scan plane quality scoring methods provide a composite quality rating (good / poor) but are unable to distinguish between poor probe position and suboptimal scan settings. Once such a composite quality score is indicated to the user, the user is typically unable to determine the next course of action - move the probe / change scan settings.

[0062] Therefore, as described herein, a quality score can be generated to guide the ultrasound operator during standard scan plane acquisition, thereby distinguishing the absence of critical anatomical structures (requiring scan plane reacquisition (e.g., probe movement)) from suboptimal scan settings (requiring setting changes to improve visibility of structures, such as gain, TGC, depth, etc.). Furthermore, the quality score can identify both the absence of critical anatomical structures and suboptimal scan settings.

[0063] This can be achieved using two neural networks: (a) Scanner Setup Sufficiency Network: predicts whether a given scan plane is abnormal due to suboptimal gain settings (oversaturation or low gain in the far field); and (b) Anatomy Sufficiency Network: predicts whether a given image is abnormal due to the absence of any critical anatomical structures. A mapping function combines the predictions from these two networks to determine the necessary actions (moving the probe and / or changing settings) to be taken by the operator in order to visualize the desired scan plane.

[0064] The present disclosure describes a method for calling out deviations (quality scores) of a given scan plane from a desired standard scan plane caused by improper probe position and / or scan settings. This quality score determines subsequent user actions - adjusting the probe position and / or changing settings (such as gain / TGC / depth settings) to obtain the desired standard scan plane.

[0065] The technical effect of the present disclosure is to provide a more descriptive and nuanced quality score than other techniques. A quality score mechanism that prompts the user to take necessary corrective actions to reach the desired scan plane will help improve the skills of inexperienced users. The method of the present disclosure can expand the accessibility of ultrasound as a modality of choice for a large number of clinicians.

[0066] An embodiment provides a method for an ultrasound system, the method comprising determining a probe position quality parameter of an ultrasound image acquired with the ultrasound system, the probe position quality parameter representing a quality level of the ultrasound image relative to a position of an ultrasound probe used to acquire the ultrasound image; determining one or more acquisition setting quality parameters of the ultrasound image, each acquisition setting quality parameter representing a corresponding quality level of the ultrasound image relative to a corresponding acquisition setting used to acquire the ultrasound image; and providing feedback to a user of the ultrasound system based on the probe position quality parameter and / or one or more acquisition setting quality parameters, the probe position quality parameter and each acquisition setting quality parameter being determined based on output from a separate image quality assessment model. In a first embodiment of the method, providing feedback to the user comprises displaying image acquisition guidance on a display device based on the probe position quality parameter and / or one or more acquisition setting quality parameters. In a second embodiment of the method, optionally including the first embodiment, displaying image acquisition guidance on a display device based on the probe position quality parameter and / or one or more acquisition setting quality parameters comprises displaying a notification that an anatomical feature is expected to be missing or has a morphological change compared to a standard scan plane based on the probe position quality parameter. In a third embodiment of the method, optionally including one or both of the first and second embodiments, displaying image acquisition guidance on a display device based on a probe position quality parameter and / or one or more acquisition setting quality parameters includes displaying a corresponding quality indicator for each acquisition setting used to acquire an ultrasound image. In a fourth embodiment of the method, optionally including one or more of each of the first to third embodiments, determining one or more acquisition setting quality parameters of an ultrasound image includes determining one or more of a depth parameter representing depth quality, a frequency parameter representing frequency quality, a TCG parameter representing time compensation gain (TCG) quality, and a gain parameter representing gain quality. In a fifth embodiment of the method, optionally including one or more of each of the first to fourth embodiments, determining a first acquisition setting quality parameter of the one or more acquisition setting quality parameters of an ultrasound image includes inputting the ultrasound image as an input to a first acquisition setting evaluation model, which is trained to output a first acquisition setting quality parameter representing the quality of the first acquisition setting. In a sixth embodiment of the method, which optionally includes one or more of each of the first to fifth embodiments, the first acquisition settings model includes an uncertainty network configured to segment the target anatomical feature, wherein the first acquisition settings model determines a first acquisition settings quality parameter by segmenting the target anatomical feature in the ultrasound image multiple times and determining an uncertainty level of the segmentation, and wherein the first acquisition settings parameter is based on the uncertainty level.In a seventh embodiment of the method that optionally includes one or more of each of the first to sixth embodiments, a first acquisition settings model is trained with a training data set comprising a plurality of training ultrasound images, each training ultrasound image being acquired at a corresponding target setting for the first acquisition setting (e.g., wherein the target setting produces high image quality relative to the first acquisition setting) and wherein at least some of the plurality of training ultrasound images are acquired at non-target settings of one or more other acquisition settings (e.g., wherein the non-target setting produces poor or less high quality images relative to the one or more other acquisition settings) and each training ultrasound image includes annotations indicating target anatomical features.

[0067] An embodiment provides a system comprising a memory storing instructions; and a processor, the processor being communicatively coupled to the memory and being configured, when executing the instructions, to: determine a probe position quality parameter of an ultrasound image acquired with an ultrasound system, the probe position quality parameter representing a quality level of the ultrasound image relative to a position of an ultrasound probe used to acquire the ultrasound image; determine one or more acquisition setting quality parameters of the ultrasound image, each acquisition setting quality parameter representing a corresponding quality level of the ultrasound image relative to a corresponding acquisition setting used to acquire the ultrasound image; and provide feedback to a user of the ultrasound system based on the probe position quality parameter and / or one or more acquisition setting quality parameters, the probe position quality parameter and each acquisition setting quality parameter being determined based on outputs from a separate image quality assessment model. In a first embodiment of the system, the memory stores an image quality assessment model, the image quality assessment model comprising an anatomical model, the anatomical model being trained to output a probe position quality parameter based on an ultrasound image. In a second embodiment of the system, optionally including the first embodiment, the memory stores an image quality assessment model, the image quality assessment model comprising one or more scan setting models, each scan setting model being trained to output a corresponding acquisition setting quality parameter based on an ultrasound image. In a third embodiment of the system, which optionally includes one or both of the first and second embodiments, the one or more scan setting models include a gain model, the gain model is trained to output a gain image quality parameter, the gain image quality parameter indicates a quality level of an ultrasound image relative to a gain setting used to acquire the ultrasound image, the gain model includes an uncertainty network, the uncertainty network is configured to segment a target anatomical feature, wherein the gain model is configured to determine the gain image quality parameter by segmenting the target anatomical feature in the ultrasound image multiple times and determining an uncertainty level of the segmentation, and wherein the gain image quality parameter is based on the uncertainty level. In a fourth embodiment of the system, which optionally includes one or more or each of the first to third embodiments, the one or more scan setting models include a depth model and / or a frequency model.

[0068] An embodiment provides a method, the method comprising determining, based on an anatomical model, that an ultrasound image includes one or more expected anatomical features; after determining that the ultrasound image includes the one or more expected anatomical features, inputting the ultrasound image into a plurality of scan setting models, each of which is trained to determine a corresponding acquisition parameter-specific image quality of the ultrasound image; and displaying image acquisition guidance based on each corresponding acquisition parameter-specific image quality of the ultrasound image on a display device. In a first embodiment of the method, displaying the image acquisition guidance based on each corresponding acquisition parameter-specific image quality of the ultrasound image comprises displaying a graphical user interface comprising a plurality of visual elements, each visual element representing a corresponding acquisition parameter-specific image quality of the ultrasound image. In a second embodiment of the method, optionally including the first embodiment, each visual element comprises a color or grayscale gradient. In a third embodiment of the method, optionally including one or both of the first and second embodiments, inputting the ultrasound image into a plurality of scan setting models comprises inputting the ultrasound image into a gain model, a depth model, and / or a frequency model. In a fourth embodiment of the method, optionally including one or more of each of the first to third embodiments, a gain model is trained to determine a gain-specific image quality of the ultrasound image, a depth model is trained to determine a depth-specific image quality of the ultrasound image, and a frequency model is trained to determine a frequency-specific image quality of the ultrasound image. In a fifth embodiment of the method, which optionally includes one or more of each of the first to fourth embodiments, the gain model, the depth model and / or the frequency model each include a corresponding uncertainty network configured to segment the target anatomical feature, and determine the image quality specific to the corresponding acquisition parameters by segmenting the target anatomical feature in the ultrasound image multiple times and determining the uncertainty level of the segmentation. In a sixth embodiment of the method, which optionally includes one or more of each of the first to fifth embodiments, the gain model, the depth model and / or the frequency model are each trained with a corresponding training data set including a plurality of training ultrasound images, each training ultrasound image of the training data set is acquired at a corresponding setting of gain, depth or frequency that produces high image quality, and each training ultrasound image includes a label indicating the target anatomical feature.

[0069] When introducing the elements of various embodiments of the present disclosure, the words "one", "a kind of" and "the" are intended to mean one or more elements present in these elements. The terms "first", "second", etc. do not represent any order, amount or importance, but are used to distinguish one element from another element. The terms "comprise", "include" and "have" are intended to be inclusive, and mean that additional elements may also exist in addition to the listed elements. As used herein, the terms "connected to", "coupled to", etc., an object (e.g., material, element, structure, member, etc.) can be connected to or coupled to another object, regardless of whether the one object is directly connected or coupled to another object, or whether there are one or more intervening objects between the one object and another object. In addition, it should be understood that the reference to "an embodiment" or "embodiment" of the present disclosure is not intended to be interpreted as excluding the existence of additional embodiments that also include the cited features.

[0070] In addition to any previously indicated modifications, numerous other variations and alternative arrangements may be devised by those skilled in the art without departing from the spirit and scope of the present description, and the appended claims are intended to cover such modifications and arrangements. Therefore, although the information has been described above in detail in conjunction with what are currently considered to be the most practical and preferred aspects, it will be apparent to those of ordinary skill in the art that many modifications may be made, including but not limited to form, function, mode of operation, and use, without departing from the principles and concepts set forth herein. Likewise, as used herein, the examples and embodiments are intended to be illustrative in all respects only and should not be construed as limiting in any way.

Claims

1. A method for an ultrasound system, the method comprising: determining a probe position quality parameter of an ultrasound image acquired using the ultrasound system, the probe position quality parameter representing a quality level of the ultrasound image relative to a position of an ultrasound probe used to acquire the ultrasound image; determining one or more acquisition setting quality parameters of the ultrasound image, each acquisition setting quality parameter representing a respective quality level of the ultrasound image relative to a respective acquisition setting used to acquire the ultrasound image; wherein determining a first acquisition settings quality parameter of the one or more acquisition settings quality parameters of the ultrasound image comprises inputting the ultrasound image as an input to a first acquisition settings evaluation model, the first acquisition settings evaluation model being trained to output a first acquisition settings quality parameter representing a quality of a first acquisition setting, wherein the first acquisition settings evaluation model comprises an uncertainty network configured to segment a target anatomical feature, and wherein the first acquisition settings evaluation model determines the first acquisition settings quality parameter by segmenting the target anatomical feature in the ultrasound image a plurality of times and determining an uncertainty level for the segmentation, and wherein the first acquisition settings parameter is based on the uncertainty level; as well as Feedback is provided to a user of the ultrasound system based on the probe position quality parameter and / or the one or more acquisition setup quality parameters, the probe position quality parameter and each acquisition setup quality parameter being determined based on output from a separate image quality assessment model. 2 . The method of claim 1 , wherein providing feedback to the user comprises displaying image acquisition guidelines on a display device based on the probe position quality parameter and / or the one or more acquisition setup quality parameters.

3. The method of claim 2, wherein displaying image acquisition guidance on the display device based on the probe position quality parameter and / or the one or more acquisition setting quality parameters comprises displaying a notification that an anatomical feature is expected to be missing or have a morphological change compared to a standard scan plane based on the probe position quality parameter.

4. The method of claim 2, wherein displaying image acquisition guidelines on the display device based on the probe position quality parameter and / or the one or more acquisition setting quality parameters comprises displaying a corresponding quality indicator for each acquisition setting used to acquire the ultrasound image.

5. The method of claim 1 , wherein determining the one or more acquisition setting quality parameters of the ultrasound image comprises determining one or more of a depth parameter representing depth quality, a frequency parameter representing frequency quality, a time compensation gain (TCG) parameter representing TCG quality, and a gain parameter representing gain quality.

6. The method of claim 1 , wherein the first acquisition settings model is trained with a training dataset comprising a plurality of training ultrasound images, each training ultrasound image being acquired at a corresponding target setting of the first acquisition settings and at least some of the plurality of training ultrasound images being acquired at non-target settings of one or more other acquisition settings, and wherein each training ultrasound image comprises annotations indicating the target anatomical features.

7. A system comprising: a memory storing instructions; and a processor, the processor being communicatively coupled to the memory and being configured, when executing the instructions, to: determining a probe position quality parameter of an ultrasound image acquired using an ultrasound system, the probe position quality parameter representing a quality level of the ultrasound image relative to a position of an ultrasound probe used to acquire the ultrasound image; determining one or more acquisition setting quality parameters for the ultrasound image, each acquisition setting quality parameter representing a respective quality level of the ultrasound image relative to a respective acquisition setting used to acquire the ultrasound image; and providing feedback to a user of the ultrasound system based on the probe position quality parameter and / or the one or more acquisition settings quality parameters, the probe position quality parameter and each acquisition settings quality parameter being determined based on outputs from a separate image quality assessment model, wherein the memory stores the image quality assessment model, the image quality assessment model comprising one or more scan setting models, each scan setting model being trained to output corresponding acquisition setting quality parameters based on the ultrasound image, wherein the one or more scan setting models include a gain model, which is trained to output a gain image quality parameter, the gain image quality parameter indicating a quality level of the ultrasound image relative to the gain setting used to acquire the ultrasound image, the gain model includes an uncertainty network, the uncertainty network is configured to segment a target anatomical feature, and wherein the gain model is configured to determine the gain image quality parameter by segmenting the target anatomical feature in the ultrasound image multiple times and determining an uncertainty level of the segmentation, and wherein the gain image quality parameter is based on the uncertainty level.

8. The system of claim 7, wherein the memory stores the image quality assessment model, the image quality assessment model comprising an anatomical model trained to output the probe position quality parameter based on the ultrasound image.

9. The system of claim 7, wherein the one or more scan setting models include a depth model and / or a frequency model.

10. A method comprising: determining, based on the anatomical model, that the ultrasound image includes one or more expected anatomical features; After determining that the ultrasound image includes the one or more expected anatomical features, inputting the ultrasound image into a plurality of scan setting models, each scan setting model being trained to determine image quality specific to a corresponding acquisition parameter of the ultrasound image; as well as displaying, on a display device, an image acquisition guideline specific to an image quality based on each corresponding acquisition parameter of the ultrasound image, wherein inputting the ultrasound image into the plurality of scan setting models comprises inputting the ultrasound image into a gain model, a depth model and / or a frequency model, The gain model, the depth model and / or the frequency model each include a corresponding uncertainty network, which is configured to segment the target anatomical feature and determine the image quality specific to the corresponding acquisition parameters by segmenting the target anatomical feature in the ultrasound image multiple times and determining the uncertainty level of the segmentation.

11. The method of claim 10, wherein displaying the image acquisition guide based on the image quality specific to each corresponding acquisition parameter of the ultrasound image comprises displaying a graphical user interface including a plurality of visual elements, each visual element representing the image quality specific to the corresponding acquisition parameter of the ultrasound image. The method of claim 11 , wherein each visual element comprises a color or grayscale gradient.

13. The method of claim 10, wherein the gain model is trained to determine a gain-specific image quality of the ultrasound image, the depth model is trained to determine a depth-specific image quality of the ultrasound image, and the frequency model is trained to determine a frequency-specific image quality of the ultrasound image.

14. The method of claim 10, wherein the gain model, the depth model and / or the frequency model are each trained using a corresponding training data set comprising a plurality of training ultrasound images, each training ultrasound image of the training data set being acquired at a corresponding setting of gain, depth or frequency that produces high image quality, and each training ultrasound image comprising an annotation indicating the target anatomical feature.

Images