The present invention relates to the field of ultrasound apparatus, systems, and methods for imaging and treatment. Specifically, the present invention relates to reducing cost, improving scalability, improving flexibility, improving computational power, of ultrasonic diagnostic or treatment apparatus.
An ultrasound imaging apparatus can be divided into two main subsystems. 1) A frontend subsystem responsible for: a) transmit signal generation and excitation of a probe; b) receiving signal, conditioning, and digital sampling; c) coordinating transmit and receive; d) send analog-to-digital convert (ADC) sampled data back to an imaging processor. A Frontend subsystem usually has a high number of physical channels, such as 128. It is common to build a 128 channel frontend system by combining a few, such as 4, PCBs together inside a customized box, where each PCB has 32 channels. The received data from these channels will be processed to construct the image. 2) Imaging processor subsystem is responsible for process the Frontend received data via beamforming, mid-end signal conditioning and processing, back-end image processing to generate the image ready for display. The most computationally intensive part of this imaging processor subsystem is beamforming. Historically for premium or highend system, beamforming is implemented by ASIC or FPGA. Diagrams of such architectures are available online. See for example, https://www.ti.com/solution/ultrasound-scanner, for various designs.
From development cost point-of-view, to develop an ultrasound imaging system with premium capability from scratch, a company or person faces the challenges of high development cost associated with building teams of 1) electronic hardware engineers to design and produce analog and digital circuits, 2) firmware engineers to drive these circuits and chips, 3) software engineers to work on various part of the system such as implementation of image reconstruction and processing algorithms, program control path, design UI, etc, 4) mechanical engineers to develop various modules of the system, such as system box/chassis and assemble the parts together.
For a company or person to develop a complete highend or premium capable ultrasound imaging system without high development cost illustrated above, there are two main approaches which are described in details below, as well as their short comings.
The first option is OEM ultrasound developed by some manufacture[6, 7], which is an ultrasound box integrated with frontend as well as hardware based beamforming. The ultrasound box will need to connect to a PC. The post processing could be located either in ultrasound box or in PC. This option works if the customer's purpose is to incorporating ultrasound image into his/her product, usually an application based on ultrasound image, without the need to develop his/her own ultrasound system. With this option, the customer cannot develop custom algorithms such as novel beamforming, since the hardware based beamforming has already been done inside the OEM box and pre-beamforming data is lost. For this case, since the post processing has already done in the OEM box or PC, there is not much can be done to develop your own ultrasound imaging system such advanced method/algorithm to improve image quality.
The second option is research ultrasound [1]-[4] developed in academia as well as industry[9][11]. These systems are for ultrasound research instead of routine diagnostic use. Research ultrasound[1]-[4][9] is an ultrasound box sitting outside of a computer and connected to the computer via high-speed links such as PCIe or USB3.0. The box either includes ultrasound front-end only which sends pre-beamformed data back to the computer [9][11], or together with beamforming module which sends beamformed data back to the computer, or with post-processing module as well which sends post-processed display ready data back to the computer. Research ultrasound provides the opportunity for a developer to design and implement his/her own specialized ultrasound system with software API to program the system.
Both the two options above have high hardware and development cost, no flexibilities in system size and shape, poor scalability of channel counts.
Besides ultrasound imaging apparatus, the current invention can also be applied to the following two cases. 1) In the case of an ultrasound treatment apparatus where the only purpose is deliver ultrasound energy into a tissue, the frontend subsystem, which drives a treatment transducer for delivering ultrasound energy, is responsible for only transmit signal generation and excitation of the probe without the implementation of receive related functionalities, such as those listed in the ultrasound imaging apparatus above includes: b) receiving signal, conditioning, and sampling; c) coordinating transmit and receive; d) send ADC sampled data back to an imaging processor. An ultrasound treatment apparatus does not have an imaging processor subsystem. 2) In the case of an ultrasound imaging as well as treatment apparatus, it includes both ultrasound imaging apparatus as well as ultrasound treatment apparatus.
The present inventor has realized the problems associated with the two approaches reviewed in the previous section for building an ultrasound system from scratch.
For the first option of OEM ultrasound developed by some manufacture[6, 7], two-box design make the system less compact and special hardware may be needed to secure and combine them together in case one combined system is desired. (The PC box can take the form factor of either a regular PC box, a laptop, or a small form factor PC). When other customized hardware is needed, it can be plugged into the PC's extension slot. This leads to a regular PC box needed (laptop and small form factor PC cannot have extension card added) which will be paired with the ultrasound box to build the complete system.
For the second option of research ultrasound [1]-[4][9][11], due to the lack of using off-the-shelf components and not optimized hardware design, such platform usually have high hardware (BOM) and manufacturing cost, bulky appearance, custom designed box and thermal management system, separated parts of data acquisition unit and data processing unit loosely connected to make the whole system. Research ultrasound [10]-[12] is using a single box design though, however, it still requires custom designed frontend PCB, custom box, and custom thermal management system, without the flexibility in choosing size and shape of the system box.
Both options above does not provide some or all of the following desired aspects for an ultrasound imaging system: 1) low hardware BOM cost, 2) low hardware development cost, 3) enough flexibilities of design of system size and shape, 4) enough scalability to build high channel count systems. Therefore, what is needed are systems and methods that overcome the above mentioned disadvantages.
Various embodiments and aspects of the inventions will be described with reference to details discussed below, and the accompanying drawings will illustrate the various embodiments. The following description and drawings are illustrative of the invention and are not to be construed as limiting the invention. Numerous specific details are described to provide a thorough understanding of various embodiments of the present invention. However, in certain instances, well-known or conventional details are not described in order to provide a concise discussion of embodiments of the present inventions.
Reference in the specification to “one embodiment” or “an embodiment” or “another embodiment” or “some embodiment” means that a particular feature, structure, or characteristic described in conjunction with the embodiment can be included in at least one embodiment of the invention. The appearances of the phrase “in one embodiment” in various places in the specification do not necessarily all refer to the same embodiment.
The present disclosure details various aspects of a new paradigm for the design and building of a complete premium capable ultrasound imaging and/or treatment system. A modularized ultrasound frontend module is designed to have a PCIe add-on card form factor which can be inserted onto any PCIe slot of a PCIe expansion subsystem to finish the hardware build of an ultrasound system. The digital data generated from Frontend is sent to the CPU/GPU memory through PCIe or other high speed connections.
In one embodiment, the PC system's PCIe expansion subsystem is the expansions slots on the PC's motherboard. The PC system's PCIe expansion subsystem box is a PC case. As shown in
In another embodiment, the PC system's PCIe expansion subsystem is the PCIe expansion slots of a plurality of PCIe expansion enclosures, where the PCIe expansion enclosures connect to said PC with high speed connections. The PC system's PCIe expansion subsystem box is the PCIe expansion enclosure case. As shown in
In another embodiment, the PC system's PCIe expansion subsystem include a combination of both the PC's motherboard with PC case and PCIe expansion enclosures.
In some embodiment, one or multiple simplified PCIe Frontend modules, which has only transmit circuit present and transmit beamforming implemented on the control FPGA, can be inserted either inside a PC case or into one or multiple separate PCIe expansion enclosures to drive a treatment transducer and hence build an ultrasound treatment system.
In some embodiment, some of the Frontend modules are connected to an imaging probe, and at the same time, some of the Frontend modules are connected to an treatment probe, hence build a combined ultrasound imaging as well as treatment system.
With low cost frontend module and usage of mass produced off-the-shelf consumer electronic components such as GPU, PC chassis/case, motherboard, memory and hard drive, power supply, PCIe enclosure, etc, the disclosed ultrasound system can simultaneously achieve previously conflicting advantages: 1 low hardware cost, 2 low development cost, 3 great flexibility in choosing system form factor with existing off-the-shelf cases and PCIe enclosures, 4 scalability to extend to high channel count systems (higher than conventional 128 or 256 channel) by using existing PCIe enclosures, 5 high computational power and easily upgradable to latest technology.
The present invention is illustrated by way of example and not limitation in the figures of the accompanying drawings in which like references indicate similar elements.
Recent introduction of GPU into the ultrasound industry has the opportunity to significantly reduce the hardware and development cost to develop an ultrasound system. As the main processing unit, compared to the alternative technologies (previous approaches) using ASIC, FPGA, specialized DSP, and CPU, GPU can reduced the cost by magnitudes due to its low hardware cost per computational power as an off-the-shelve consumer product and extremely low development cost as only GPGPU programming is needed. For example, the ultrasound beamformer can be implemented by using off-the-shelve consumer grade GPU with magnitude of reduction in hardware cost compare with industry grade ASIC, FPGA, DSP, not to mention that the reduction of development cost is even more. It opens up extremely fast approaches to develop new ways to implement beamforming, image reconstruction, new imaging mode, other innovations, etc. It also provides a convenient low cost way to extend system's computational power by adding additional GPU cards, or simply upgrade to the latest GPU model.
The high data speed of PCIe extension bus makes the transfer of the RF pre-beamformed data into CPU or GPU memory become feasible. For example, a Frontend module with 64 channels with 64 MHz sample rate ADC and sample size of 16 bits will generate 8 GB data per second. This high data rate can be achieved by, for example, PCIe 3.0 x8.
Recent advances of the ultrasound frontend integrated circuits (IC) with highly integrated off-the-shelf frontend chips for transmit, receive, and analog to digital converter (ADC) make the development of compact ultrasound frontend module within a small form factor such as a PCIe card become feasible. The prevalence of PCIe IP controller on low end FPGA makes the low cost frontend card possible. Now it is practical to produce low cost, highly integrated frontend module in a form factor of a PC extension card.
The present inventor has 1) observed above evolutions; 2) realized that to build a premium capable ultrasound system, the only parts needs to be built is the Frontend: Image Processor and other peripheral subsystems can be assembled via off-the-shelve components; and 3) invented a modularized way to develop ultrasound system with low hardware and development cost, as well as high scalability of physical channels and computational power.
To achieve these improvements, an ultrasound frontend module is invented with a PCIe extension card form factor which can be inserted into a motherboard PCIe slots, sit inside and secured to the PC case without design and manufacture of custom system box and thermal management system. One or multiple frontend module can work together as a frontend subsystem to achieve various physical channel configurations. An ultrasound probe can be connected directly to the connector sitting on the frontend card's PCI bracket, or through a probe connector adaptor, especially when multiple frontend cards are involved.
With modularly designed system under the standard PC high speed expansion bus (PCIe) architect, current invention provides a fast and convenient way to develop ultrasound imaging systems with: flexibility of choosing system form factor, size, and shape; scalability to extend to higher channel counts; lowest cost by maximum using off-the-shelve consumer components; best computational power by free choice of the latest GPU model and number of GPUs.
Using a standard or customized PC case as system box has following advantages. 1) As an off-the-shelf component, it has low hardware cost and zero development cost. 2) It is flexible to choose the size and shape of the system, as well as other components such as mother board, power supply, etc. Closed shielding enclosure is used for the frontend card to avoid environmental EM interferences inside of PC case. The cooling fans of the PC chassis/case's thermal system and the cooling fans/heat sink on the Frontend module work as the thermal management system to keep the frontend modules and the whole system from overheat. The ultrasound pre-beamformed data collected by the frontend will be transferred to CPU or GPU memory through PCIe bus. Power regulators are used to convert the 12V, 5V, and other power output from the power supply unit (PSU) into various analog and digital voltage rails used by the frontend module. These power regulators can be located either on each frontend module or on a separate Frontend Power Supply Module (FPSM) which also has a PCIe form factor and could be access and controlled by CPU.
The idea of PC based ultrasound was first patented in [5], where a workstation or PC's CPU(s) is(are) used to perform the most of the image processor's work to get the image, instead of conventionally a series of serially connected special purpose circuits (made out of ASIC or FPGA), including beamforming, midend signal processing, and backend image processing, at the time of the invention around 1990s. One illustration of PC based ultrasound is shown in
U.S. Pat. No. 8,824,743 [8] has mentioned an implementation of Frontend circuit as “plug-in module” which connect to the PC's expansion slot, as illustrated in FIG. 6 of [8] but it has several important differences with the proposed invention. The author of [8] is advocating the “separate ultrasound data collection box design” and teaches away from the current invention. See [9] for more information about the Vantage system. The “plug-in module” disclosed in [8] is the Vantage Data Acquisition system shown as the big box in the left next to the host computer in Figure “Verasonics Vantage Research Ultrasound System” of [9]. Below are the list of differences: 1) The “plug-in module” custom hardware is preferably housed in a custom enclosure, which is an ultrasound frontend box sitting outside of a computer and connect to the computer with high speed links. 2) A large block of expansion memory is used as part of the custom “plug-in module” hardware to store the RF data from the frontend circuit. Since this expansion memory on the “plug-in module” is accessed by the CPU during the “pixel-based reconstruction” process, it needs to store the RF data composed of an image frame. According to the calculation in [8], each transmit/receive event will use 1 MB data for a 128 channel system, and a typical image RF data from 256 events will need 256 MB memory to store. This large block of extra expansion memory is not needed in the proposed invention's frontend module, because the 256 transmit/receive event's data is send to PC system's memory or GPU's memory immediately after each transmit/receive event without accumulating all of them, hence 1 MB data memory is enough which corresponding to a single transmit/receive event. Compare to [8], the proposed invention's frontend modules need to store as low as 1 MB data corresponding to a single transmit/receive event, which is two magnitudes reduction from [8]'s 256 MB. For a typical system with 4 frontend modules to compose a 128 channel system, each frontend module only need 0.25 MB storage for the data. This can be easily accommodated by current lower end FPGA's internal logic or a small SRAM outside of FPGA. Hence reduces cost. 3) The space inside a PC box is a strong electromagnetic (EM) radiation environment. To protect the analog circuits in the frontend, shielding box is needed to void noise/artifacts in the collected data due to the EM interference. So shielding box on PCB is preferred to the design of the current invention where the frontend PCB needs to sit inside PC box. 4) [8]'s “plug-in module” requires an external enclosure to host multiple PCBs with each PCB is has a fixed number of physical channels. This requires design and build extra circuits inside the enclosure to aggregate the data from each PCB and send the aggregated data back to PC. However, the current invention will not require this aggregation circuit since each individual Frontend module will send its data to PC or GPU memory directly and independently. Hence significantly simplifies the architecture and reduces cost. 5) [8] doesn't include the clock and transmit/receive event synchronization signal on each frontend module with off-the-shelve connectors which enables the building of a system with scalable number of Frontend modules without the need of building a customized external enclosure since in [8] the external synchronization signal is on the external enclosure. Note the “plug-in module” in [8] is different from the frontend module in the current invention with above 5 differences. The current invention provides reduced system cost, reduced development cost, and better flexibility in building customized systems compare to [8]. Note that in another embodiment of the current invention, where very high channel count requires multiple boxes to hold a large number of frontend modules, the frontend module design enables the use of existing off-the-shelve PCIe enclosure to hold multiple frontend modules and send the aggregated data back to PC system, this approach still has the above differences 2)˜4) compared with [8]. Although this embodiment also using external box, these box are off-the-shelve PCIe enclosure and hence reduced development cost associated with the customized enclosure in [8].
To meet the requirements of medical device regulation, medical degree off-the-shelve power supply unit can be used for this purpose and avoid delay to the market.
The invented platform also enables a new market: individuals, who do not have the resources to develop a complete ultrasound system, can develop their own state-of-the-art ultrasound system by plugging in frontend cards into a PC system and connect an ultrasound probe. The standard accessories and modularized hardware and software enables developers to develop or customize their own systems by off-the-shelve components. There is significant flexibility of the system developed by using the disclosed platform. For example, when multiple frontend cards are used, the clock and trigger signals from the master needs to be connected to other slaves, which can be solved by using off-the-shelve Y cables. With PCIe card form factor based frontend module as off-the-shelf component, developer of ultrasound system can simply choose a few frontend cards, two for example, plug them into the existing PC case, as well as GPU cards for high computation power and good system performance, to finish up the hardware development of the system.
Another embodiment of the current invention is building high channel count system by using off-the-shelve PCIe enclosures. One PCIe enclosure can be used to hold multiple frontend modules, hence is called a high channel count module (HCCM). One PCIe enclosure can also be used to hold multiple GPUs. To scale the system in both physical channels and computational power, the system just need to increase the number of HCCM and GPU enclosures.
It will be apparent to those skilled in the art that various modifications and variations can be made to the disclosed embodiments. It is intended that the specification and examples be considered as exemplary only, with a true scope of the disclosure being indicated by the following claims and their equivalents.
Provisional Application: Application No. 63/094,897
| Number | Date | Country | |
|---|---|---|---|
| 63094897 | Oct 2020 | US |