Imaging The World – A System Based Solution to Imaging in Resource Limited Counties - HD

Introduction

My name is Dr. Brian Asca.
I'm an employee of the US Department of Veterans Affairs, and also of the Food and Drug Administration.
I'm gonna be talking to you today about a global outreach program that we have developed.
The title of the organization is called Imaging the World.
I'll be talking to you about how a systems based approach can be used to develop an imaging program in other countries.

The Global Imaging Gap

There's an imaging gap that exists in this world.
Through the 1960s, medical diagnosis was performed the same worldwide.
You would do a physical and history and physical examination.
You would arrive at a preliminary differential diagnosis, then you would obtain lab tests and perhaps x-ray to confirm.
Since then, developed countries have come to rely heavily on cross-sectional imaging.
They use this to even augment and replace the physical examination, and definitely to shorten the differential diagnosis.
Unfortunately, developing countries have not participated in this imaging revolution.
They've never had the resources to develop cross-sectional imaging.

Reasons for the Gap

Why is that?
Why did they not participate in this revolution?
A lot of reasons that are pretty obvious.
There's political instability, financial instability in these countries, which prevent them from growing the infrastructure of their healthcare system.
They have limited personnel, limited educational facilities, limited electrical power, limited money, and there's problems with shielding and construction, and also with their communications infrastructure, which is far less advanced.

Okay, as a result of that imaging and resource limited environments, such as developing countries or rural areas even of developed countries, is mostly limited to x-ray.
Image-based diagnosis is really not used and is almost non-existent.
Again, the diagnosis primarily relies on history and physical examination.
Advanced imaging is, if they have, it is limited to ultrasound.
They have almost no CT MRI nuclear medicine or PET scanning.

Systems Issues in Developing Countries

So we have image-based diagnosis and treatment, and if we treat it as a system, we have some systems issues in developing countries.
The first one is lack of access.
They have to have an imaging system to use it.
The second is a shortage of technologists to run that system.
The third is a shortage of imaging interpreters, people who can interpret the images because there's no experience developed in that country.
There's also a lack of facilities for definitive care.
If you make the diagnosis, can you treat that diagnosis somewhere?
And also, there's limited patient familiarity and trust in imaging since they've never experienced this before.
Of course, since they don't have an imaging system like that, they don't have an appropriate quality monitoring system.
And if they had the imaging system, there would be problems with sustainability and scalability of the system in that country.

Founding of Imaging the World

Imaging the World was founded in 2008 to address these issues.
These issues are common to most limited resource environments, even in the United States, as I mentioned earlier.
So let's tackle these issues one at a time.

Lack of Imaging Systems

The first one is lack of imaging systems.
The solution is to distribute small ultrasound systems widely to villages and small clinics.
The reason we don't use CT MRI or PET scanning is because they're expensive and require a lot of infrastructure.
Ultrasound does not.
These ultrasound machines that can be distributed can range from between $2,100 to $3,500.
Ultrasound is safe, so there's no shielding required, and a low level of support and infrastructure is needed.
These can run on as low as 75 watts of power.
Our goal in Imaging the World was to place one machine within one hour for each person being served using their normal mode of transport.
For instance, they may have small motor scooters, or they may have to walk.

Low-Cost Ultrasound Scanners

So let's talk a little bit about ultrasound scanners that are low cost and designed for low resource environments.
Vendors are aiming at this large and growing market, even though the per unit cost is very low, the number of units that could be potentially sold is very large.
These machines typically are tailored to produce low costs and declining costs.
They have a certain performance level, which is gradually increasing.
They have a limited feature set, which is also gradually increasing.
They generally have low power consumption, and they leverage existing computer technology and tablet technology.
Here's a couple of examples of various types of lower cost machines.
A few years ago, we used the Terra on T 3000, and it kind of lists for $40,000, which is not really low cost.
More recently we have the Vell D system, which comes in at $1,900 US, and the most recent edition is the Phillips Lumify, which has really high performance and comes in at less than $15,000.

Lack of Trained Sonographers

The next issue is lack of trained sonographers.
While ultrasound machines are cheap, sonography traditionally has required considerable training.
Training of sonographers is expensive and time consuming, taking six months to two years, the sonographer must have training in both ultrasound anatomy, how to operate the machine and in pathology, since they select the images that are saved, and that are interpreted by the radiologist or other imaging professional.
The solution is to develop new imaging protocols that require no knowledge of pathology and internal anatomy, and therefore can be trained.
The people can be trained in this very rapidly.
Our ITW scan protocols are like this.
I like to say they're so simple. A child can learn them.
Certainly any adult person can learn them.
Familiarity of the computers is helpful.
Their volume imaging dependent, meaning that you sweep across an area of interest and the scanner is capturing images at a certain rate.
And then in the end, you have a bunch of images that cover that area thoroughly.
These sweeps, imaging sweeps, usually five to six sweeps are performed using only surface landmarks for reference.
And those are easily taught.
The sweeps are selected based on experience to maximize diagnostic information.
Ultrasound scanner presets are used to automate machine adjustments and image labeling.

So far, Imaging the World has created thyroid, gallbladder, kidney, pelvis, limited OB, and more recently breast, a scan for fluid in the abdomen in emergency situations and even a bone survey type scan.

Training Methods

How's the training performed?
We try to have the people training to learn to watch the patient, not the scanner.
They want to be sure their transducer is moving across the patient in a nice, smooth, regular fashion.
And although the scanner has a screen, since they are not being trained to interpret that screen image, they're, they should watch the patient instead to make sure the scan is being performed properly.
We found in our training that color posters are a very useful addition.
The poster shows how the probe should be held, how it should be swept across the patient, and how many different types of sweeps should be done in and in what locations.
On this slide, you can see the example of the posters.
They're very handy to taped to the wall near the patient in case the person performing the scan starts to forget how to perform portions of a scan.

Examples of Volume Imaging

Here's a couple of examples of volume imaging.
The one on your left hand slide is a longitudinal sweep across the pelvis, and you can see on that sweep as it starts, there's a cyst in one ovary and there's the uterus.
And then the other ovary was on the other side.
Again, the cyst uterus and no other masses or fluid collections.
On the right hand side, you see a kidney ultrasound.
The kidney is outlined here, we see the entire kidney from one side to another.
We can determine if there's a mass or hydronephrosis in that kidney as well as stones.

Here's an obstetrical ultrasound scan performed from the low portion of the patient to the high portion of the patient starting just above the pubic bone.
As you can see, the head is down because the first images are of the head.
And as we climb higher, we see the liver.
And the kidney showed up right there very briefly.
We also can see the placenta lying posterior and to the left hand side and can see clearly that the placenta is not between the head and the lower portion of the uterus, thus showing that there's no evidence for a placenta previa.
These scans, although appearing to be very limited, often show enough fetal anatomy to help rule out fetal abnormalities.

Testing Protocols

So these protocols were designed and they were tested first in Belize.
And then more recently in an image quality study in Uganda in 2016, volunteer trainees were trained over a three day period, and then were tested by conducting 300 scans on nine normal volunteers.
The image quality and organs coverage were examined by expert reviewers and found to be adequate.
More recently, an observer agreement study, which is now continuing to be underway, is performed to evaluate the level of agreement between interpretations from ITW type scans interpreted by six readers to a traditional scan obtained at the same time by an experienced sonographer.
In a subset analysis, there is approximately 80 to 90% agreement.

Lack of Trained Interpreters

The next issue is the lack of trained interpreters.
We've covered how the operators are trained.
Now, what about the people interpreting these scans in rural areas?
There's often a very severe shortage of radiologists or other professionals who can interpret the ultrasound examination.
The solution to this is to move the images where trained professionals are able to interpret the studies.
The way we do this is send the images to an international image server, which is located in Arizona.
Qualified volunteers can log on and interpret remotely performed scans after being properly credentialed.
Structured reports are sent electronically back to the hospital via email and to the clinic operator via a text message.

Image Flow and Data Flow

Here's an example of the image flow and data flow.
In this system, you have the ultrasound scanner, and then it communicates with a small netbook computer using DICOM and a wireless ethernet connection.
In the netbook storage, arranging the images for transmission and compression of the images for transmission is performed.
Here, the transmission is transmitted to a cell phone network to the internet, to the receiver at the site of the PACS system, or storage in Arizona, where the images are decompressed converted back into DICOM and then sent to a McKesson PACS for display reporting by volunteers.
The final step is to using a system called PeerView send the reports by email to the referral hospital, so they're available in case the patient gets sent to the hospital and also as a text message to the cell phone of the person operating the scanner in the clinic.

Here's an example of the transmission setup.
This is at Ngo in Uganda.
Here's the scanner.
There's the netbook that does the transmission.
This wire goes to a antenna on the outside that boosts the signal for transmission to the very limited cell phone capability they have in this region.
The wireless router over here is used to take the signal from the ultrasound scanner and transmit it to the netbook.

Image Viewing and Reporting System

The image viewing system looks like this.
This is a dual monitor system, although you can use a single monitor, four monitors or whatever you have.
The viewing system software is donated by McKesson.
Their software can run on almost any computer, even a netbook.
The servers hosted in Arizona, but there's worldwide access via web-based viewers, which is this an example of.
And there's an integrated reporting system.
So this is the work list of patients for which interpretations must be performed.
Each one of these little windows has a sweep in it, and you can simply click on it with your mouse and scroll through it very much like a set of CT images.

Here's an example of a simple workstation that we hand carried to Uganda.
This is located at Kali Mission Hospital, which is where the images from Ngo would be sent for interpretation or for reviewing prior to treatment of the patient.
So we bring a small laptop and a second monitor so that the text can be shown in this monitor, and the images can be looked at on this monitor.
The reporting system is generated by check boxes.
The first thing the interpreter does, or the last thing they do, is to categorize the findings that they see on the study in order of importance.
So the first two zero and one are normal.
The first one zero means that the studies of poor quality and additional images need to be obtained.
One and two are findings that are relatively normal and not of concern.
Three is probably normal, but an additional scan may need to be done.
Four and five are positive findings that need to be treated.
A five is a severe finding that the patient must be sent immediately for more definitive healthcare.

This is an example for an OB study.
So clicking on the positive opens another set of check boxes for each area for which we provide an interpretation.
On this image, you can see that the fetal position, we have several choices head down.
The rear end down the fetus could be in transverse position, or it could be moving around and be variable.
Each subset has a bunch of choices that the person clicks on, and that the system automatically generates a report based on those responses.

Data Transmission and Compression

So the next step is transmission of the image data over what is a very slow network.
The transfer of speed to the internet is extremely slow and may be unreliable as well.
In parts of rural Uganda, it's only 12 kilobits a second, which is similar to what we used to use for dial up modems back in the 1970s and 1980s.
In the United States Volume Imaging TR requires transfer of large number of images, typically 500 to 800 images per study.
But the solution to this is to exploit the fact that images are part of sweeps, and they're basically short videos, and there's a lot of fancy, high compression video technology available with which to compress these images.
This has been developed largely by the movie industry and video industry to compress data for transmission over satellites.

Another issue is you don't have any signal.
The signal can drop.
We all experience this even in the United States where our cell phone signal drops out or it fades out.
And one solution to this is to provide a bigger antenna.
Here you can see at NGO clinic, we fastened a large antenna, a yagi array to a sapling that was cut down for this purpose.
And shortly, these village volunteers are gonna raise this pole over the clinic to provide increased signal.
Another option that we've turned to more often is using a electrical signal booster.

So how is the data compression performed?
We use MPEG four, what we call MPEG four level two, and we use freeware to do this.
The freeware is called XviD, and that's the compression scheme as well.
We're able to achieve 48 to one compression, or sometimes even higher than that, while maintaining high image quality.
The entire sweep, which is 40 megabytes to 80 megabytes, can be compressed down to the size of a single image, typically 360 to 500 kilobytes.
The transmission over a slow net, even a slow network, only takes about five seconds.

Here's an example of the type of compression we apply and the image quality.
On the left hand side, you see the image at original quality, and on the right hand side, you see this same image after having been compressed by about approximately 40 to one.
And you can see there are almost no perceptible difference.
This image is a little bit smoother with a few less dots in it, then the original image.
But actually some interpreters like this image better.

How does it compare to what's used traditionally in medicine?
Here you can see our image quality after compression at approximately 40 to one.
And this is what it looks like if you compress using traditional compression technology used in healthcare.
As you can see, the kidney shown here is not even visible, and this does is not even recognizable as an ultrasound image.

Lack of Facilities for Definitive Diagnosis and Treatment

The next issue is lack of facilities for definitive diagnosis and treatment.
Our solution to this is to upgrade the ultrasound capability at the local hospital.
So in this case, we get new equipment either donated or purchased.
We have a trained sonographer.
In the case in Uganda, there's a place where the sonographer can be trained, called Ernest Cook Ultrasound Research and Education Institute.
And this can be done in conjunction with the World Federation of Ultrasound in Medicine and Biology program.
This training is performed, it typically takes about six months, and you can do it combining e-learning with hands-on training.
So that upgrades the imaging capability of the local hospital.
But we, in addition, need to upgrade the treatment capacity of the local hospital.
This can be done by partnering with another organization, say a volunteer organization of surgeons, or we can identify and recruit other treatment facilities to help manage these patients that we identify as being abnormal.

Here's an example At Kali Mission Hospital, both a new ultrasound machine and training for the sonographer were required.
Here's the ultrasound suite, which is basically a single room with a very old ultrasound machine.

Future of Ultrasound Training

What about the future of ultrasound training?
Well, one potential idea for training is to have more advanced simulations so people can practice the sweeps and ultrasound image creation on a more realistic environment.
Here's an example of an ultrasound transducer whose position is determined by GPS technology, there's a pressure sensitive membrane that acts as the patient, a simulation of the patient's skin, so the operator can learn how much pressure to apply and uses technology called a nodus and gyro.
So it can detect the membrane form deformation, simulating pressure on the skin.
And this provides an inexpensive solution to determining whether they're scanning in the right direction with the proper pressure and the right transducer angulation.

Another important aspect of training is to have an international site for people to go to, to refer for additional information on ultrasound technology.
Having training, support and also diagnostic support.
So Imaging the World is such a site and has been extremely helpful as an educational site for people in developing countries.

Limited Patient Familiarity and Trust in Medical Imaging

An additional issue is the limited patient familiarity and trust in medical imaging.
Many of these people would often go to their local witch doctor or local healthcare provider, who may have had no training or very limited training in modern medicine, they're not used to imaging and they view oftentimes view it with great suspicion.
The solution is to start training these people to understand what medical imaging is and how important it can be to their health.
Oftentimes, this is done by word of mouth, sometimes by local television and radio news.
One can also do a marketing campaign.
In Africa, a lot of information is transmitted by songs and dancing.
So in ITW, we've arranged for some of the villagers to create new dances to help educate their peers in the village about these programs.
In addition, we can combine other health related activities in these training programs such as the need for immunization and standard basic health and hygiene education.

Quality Assurance

An important issue in any imaging system is quality assurance.
We have to be sure that the images are created properly, the image quality is high enough, the interpretations are provided in a timely fashion, and that the quality of the interpretations is adequate.
The solutions that we have provided are operator checks and peer review using our integrated reporting system in the stored PACS images.
Because all the images are stored in a local central server, they're always available for review and secondary review to determine if interpretations are accurate, if the image quality is appropriate.
And by looking at the images, one can tell if the images were acquired in an appropriate fashion.
If there a problem is identified, corrective training can be provided using emails, posted videos tailored to the specific problems or providing additional reference materials to the person.
Rarely, a person may need additional hands-on training.
The ultrasound system software must be renewed periodically, and this prevents abuse of the system.
Occasionally, systems will be stolen and be used to scan people for profit, which is not the mission of the ITW, and but the system software will expire, and then the ultrasound system will no longer be usable.

Sustainability and Scalability

Sustainability and scalability are very important.
This is a work in progress.
There's always challenges.
One advantage is we have is when riding the downward cost of computer technology constantly, the machines are getting cheaper, the ultrasound transducers are getting less expensive, and the demands for resources such as power and training and equipment are gradually falling.
Another way that you can obtain sustainability and scalability to partner with governments to hand over the program.
Our program is designed so that we train new trainers in country and after a short period of time, they begin to do training, freeing up Imaging the World personnel to move on to a different country or a different region of a country.
Again, it's very important to have active participation of the local healthcare community to give them ownership of this program.
Without ownership, ITW would be trapped performing these operations on a continuing basis, which is not sustainable.
It's important to have participation of the radiologists in country and the technologists to help support interpretations.
Overseas interpretation can be performed, but it's more useful as a backup to involvement of the local healthcare providers to provide readings on the scans that are obtained.
Constant readjustment and realignment with changing healthcare priorities and needs are also required.

Summary

So in summary, Imaging the World has developed a systems based imaging model designed to minimize the use of resources and allowing much more rapid deployment in Uganda and other kind countries.
And active participation by all the partners involved has allowed our initial efforts to be successful.
But many challenges lie ahead with execution, and but we are optimistic about bringing this advanced medical imaging to rural areas everywhere in the world.
Thank you very much.