A new study explores novel technology for COVID-19 diagnosis using ultrasound.

From The University of Pennsylvania,

A new study explores novel technology for COVID-19 diagnosis using ultrasound.

Penn’s ultrasound research team introduced a novel technology that detect lung pleura (membrane) changes related to COVID-19 with high accuracy. Lung ultrasound has emerged as an ideal imaging tool for COVID‐19 diagnosis because of its high sensitivity, safety, portability, and affordability. However, a significant disadvantage is that it is highly user dependent, and not all clinicians have training in performing lung ultrasound and reading the images.  To overcome these limitations, the team explored a new technology that measures the thickness and shape of lung pleural. This technology showed high diagnostic accuracy compared to traditional lung ultrasound interpretations. The technology when implemented on ultrasound devices that are currently used in clinical practice can be of great value in improving diagnosis and monitoring of COVID-19 patients.


4.5 8 votes
Article Rating
Newest Most Voted
Inline Feedbacks
View all comments
April 23, 2021 10:27 pm

I am sure this kind of image processing perfectly fits AI capabilities.
A human doctor should be excluded.

Joel O'Bryan
Reply to  Alex
April 23, 2021 11:44 pm

Actually properly trained human eyes and brains can detect much more than any current level of computer vision-AI.
The problem of course is that it takes years of training and experience to make a good radiologist. But a good radiologist or experienced Internal Med doc reading and ultrasound costs money, where AI’s can be propagated once trained like installing a program.

Last edited 1 year ago by Joel O’Bryan
dodgy geezer
Reply to  Joel O'Bryan
April 24, 2021 1:09 am

Machines should be better than humans at any specific precisely-definable task. However, humans are general-purpose machines, and can bring multiple types of understanding into play, including indefinable techniques such as intuition…

Reply to  Alex
April 24, 2021 9:00 am

It isn’t just reading the image but the process of actually doing the ultrasound itself. In fact, the same person can do multiple ultrasounds on the same person and get slightly different results. A slight change in position makes a significant change in the image. The ambiguity you end up with is something that no AI can account for.

Reply to  StevenF
April 24, 2021 5:27 pm

If I’m understanding you correctly, there is ambiguity in reading ultrasounds because of variability in how the ultrasound is administered. If that’s the case, why are people any better at dealing with that ambiguity than AI? AIs (which in this case really means machine learning, not “strong” AI) can be trained on images just like people can, in fact you can probably throw a lot more training data at the machine learning system than any one human can absorb in their training, including known ambiguous scans–in which case it should be able to distinguish positive, negative as well as ambiguous cases.

Reply to  Alex
April 24, 2021 10:27 am

That’s the kind of thinking that created the covid disaster.

April 23, 2021 11:37 pm

It sounds like this technique has advantages over x-rays, CT scans, and MRI. On the other hand, all forms of imaging are of limited use in diagnosing and treating Covid-19. link

As far as I can tell imaging, of all sorts, does not significantly improve the diagnosis and treatment of Covid-19, but I am not a medical doctor.

April 23, 2021 11:47 pm

This is an interesting technical approach that seems to be of limited availability outside the wealthy countries. Valuable in cases of admittance to hospitals for directing treatment of actual WuhanVirus cases; & of course, where it can be coded for medical insurance to pay for it. The hospitals can bill a high amount, the insurance will authorize payment of just about 60% & the hospital can then write off the billing difference not paid as a loss for tax purposes.

Jeff Meyer
Reply to  Scissor
April 24, 2021 8:05 am

I guess we are following in Rome’s foot steps, only quicker.

April 23, 2021 11:53 pm

Can it differentiate between COVID-19 and other respiratory infections? Influenza A or B, other corona viruses?

The Dark Lord
Reply to  MarkH
April 24, 2021 1:15 am

of course not … and anyone sick enough to pop on this test CAN TELL THE DOCTOR heh my lungs aren’t right …
good lord … its like poking a dead person with a needle to surmise they are dead … useless after the fact …

Last edited 1 year ago by The Dark Lord
Fred Streeter
Reply to  The Dark Lord
April 24, 2021 6:04 am

This diagnostic test, should it ever see the light of day, is to determine, as early in the infection as possible, whether the patient has contracted CoViD19.

Of course you can tell your Doctor that your lungs “aren’t right”, but that does not mean you are about to die. You may have, say, bacterial pneumonia.

Being shunted off to a Covid Ward would be the last thing you would want.
And, probably, the last thing you would get.

Reply to  Fred Streeter
April 24, 2021 3:26 pm

It’s a diagnostic test for pleura, not COVID. One of the most frustrating thing in this whole pandemic has been to explain to people that this is a common symptom of flu-like infection and in no way COVID specific.

Why do people insist on trying to make COVID into a scary, do they not realize it is their fear that has been killing people? Willful ignorance does come with a cost.

Reply to  Beeze
April 24, 2021 4:38 pm

I have come to the conclusion that many people LIKE living in terror.

Fred Streeter
Reply to  TonyG
April 25, 2021 1:02 am

So far, I have not met anyone who was terrified by Covid.

Careful to mask-up, distance, use hand sanitizer, etc.,?Yes. But “terrified”? No.

Reply to  Fred Streeter
April 25, 2021 1:45 pm
Reply to  TonyG
April 25, 2021 1:55 pm

Due to lacking context in the email updates: the tweet I posted with “here ya go” was for Fred Streeter

Fred Streeter
Reply to  Beeze
April 25, 2021 12:46 am

It’s a diagnostic test for pleura, not COVID.

“Quantitative pleural line characterization outperforms traditional lung texture ultrasound features in detection of COVID‐19.”

You were saying?

Fred Streeter
Reply to  Beeze
April 25, 2021 12:55 am

I leave scaring people to the media.

You have taken my flippant Covid Ward remark too seriously.
Lighten up.

April 24, 2021 1:41 am

The original paper is poorly presented with no description of the cases vs controls..Apparently 7 pts with CoViD diagnosed by (?)independent means were US’d and compared to 13 (?)healthy controls—They found discernable evidence of thickened pleura in 6 of the 7 cases compared to controls. …No comparisons to pts with other respiratory/pulmonarypathology…This should have been published as a letter-to-the-editor at best, with call for further investigation.

Reply to  guidoLaMoto
April 24, 2021 4:12 am

I clicked through to the study (a short read, I encourage commenters to take a look) and the fact that they only identified 6 out of 7 cases (86% or 14% failure rate) jumps out.

OTOH, the PCR test, according to the fellow who developed the test, should not be used to diagnose Covid-19. Something fast and accurate is needed.

So as I understand it, the PCR test generates a huge number of false positives (50% or higher? I can’t recall), whereas this method, using ultrasound, had ‘only’ a 14% failure rate. And this is in an early trial where no one is really an expert with the technique just yet.

I have to agree with you, guidoLaMoto, that a “call for further investigation” is in order. It does show some promise if the technique can distinguish Covid-19 from all other types of lung disorders. (I took that as your main point, guido; it was not demonstrated that it can.) Otherwise… give the researchers a “nice try, fellas.”

Ultrasounds are quick, so it would be a fast, more reliable diagnosis of Covid-19, if it truly is Covid-19 specific, when someone presents at a hospital with symptoms.

Unfortunately, the current financial incentives to hospitals are to diagnose Covid-19, so perhaps a fast, more accurate test is a bug, not a feature. But that’s another topic altogether. I’m not going there.

Fred Streeter
Reply to  H.R.
April 24, 2021 6:22 am

I too, took this paper to be a “call for investigation”.
The study appeared to me to be the best that they could do with the resources at hand.

Thanks for your post.

Reply to  H.R.
April 24, 2021 9:05 am

Where did you hear that PCR has a high percent of false positives? Antigens testing has a high false positive rate, but a properly performed PCR is pretty low. The sensitivity and specificity run in the high 90s, close to 100%.

Reply to  StevenF
April 24, 2021 2:01 pm

My understanding was that the PCR test was about 97% sensitivity and similar on specificity (at least a used in Aus. Different Ct values could greatly effect the accuracy).

Even with that, I was interested in how accurate it actually would be in practice so I calculated the Positive Predictive Value for it. Given that at the time the number of positive results was running steady at about 0.3%, using an assumed prevalence of 0.3% leads to a PPV of something around 11%. Even if the prevalence was 3%, PPV would have only been ~50%.

What would be needed is to use an initial positive result as a screening test. Then repeat the test independently. The combined tests having a sensitivity (for two positive results) of 99.9%. Two independent positive PCR tests would have an effective PPV of >90%.

My understanding though is that a single positive test result was always taken as confirmatory and even if follow up tests came back negative that “case” remained in the statistics.

There seems to be so much dirt in the statistics that it will be almost impossible to get any actual useful information out of it. It’s hard to compare to previous epidemics because everything has been redefine; “case” is now a positive test result, “death” is attributed to the disease very loosely.

Discussion of issues around COVID has been stifled with any scepticism of the official narrative effectively outlawed on main stream media/ social media.

I tend to put more weight in the arguments of the people speaking out. Particularly the senior doctors and researchers. These people have everything to lose and nothing to gain by speaking out. The people pushing the “scientific consensus” line, however, have no skin in the game. Vaccine companies have immunity deals. Pro-lockdown academics are shielded within their positions from the real world devastating effects. Although, if it does turn out that the vaccines are not as safe as they are assumed to be and end up causing mass deaths or infertility etc. No immunity deal will save them.

Reply to  StevenF
April 24, 2021 6:07 pm

Depending on how many cycles you run it for, PCR tests can have anywhere from okay to zero diagnostic value. It picks up viral fragments, so it can’t tell you if someone actually has an active infection with live virus, and if you run it with enough cycles you can end up identifying random bits of genetic code.

Even Fauci pointed out that at the number of cycles many labs are using (and not reporting) you might be better off using a dousing stick, and cycle numbers are not routinely reported. It’s completely unscientific from top to bottom.

Granum Salis
Reply to  Beeze
April 24, 2021 8:36 pm

I agree with the first 1 and a half sentences of your post.

I’m not aware that increasing the number of cycles introduces randomness into the sequence that is being amplified. It seems possible that it might.

Obviously, the diagnostic validity is diminished with the cycle number but, epidemiologically, a positive at high CT should still be of interest with regard to prevalence.

Reply to  Granum Salis
April 24, 2021 10:32 pm

It depends how specific the actual strands being identified are. In the end the viral genome is just a collection of codons and amino acids that are not specific to any organism. My understanding (which may be mistaken) is that the more cycles you run, the bigger the chance that you are just finding smaller puzzle pieces that just happen to fit into the target sequence.

Reply to  Beeze
April 26, 2021 10:16 am

That was my understanding too, Beeze. And I might well be mistaken, too.

If run at around 23 or fewer cycles, it was a strong indicator of Covid-19 and you definitely were sick with a corona virus. But many testing centers were running it at 30 or 40+ cycles.

It’s hard to say how many cycles were being run though, because for quite a while, I’d read comments from people who were trying to find out how many cycles were being run and it seemed to be a State Secret.

Again, the financial incentives were to find Covid-19, and overclocking the PCR cycles was pretty much guaranteed to find something.

The other oddity was that hardly anyone got influenza A or B in 2020. I think in June or July last year there were only a dozen or so cases of each recorded on the CDC site. You may have seen jokes in comments that “Covid-19 cured the flu.” I was seeing them.

April 24, 2021 4:12 am

recent usound was excellent and showed a lot more than I wanted to have found
follow up CT scan was parlous in comparison, scant detail
if you removed name from both you’d have not thought they were the same body examined
CT scans at far higher cost and chances of after effects than u sound as well

April 24, 2021 5:52 am

So what if you get diagnosed positive? Here where I live (a few miles from two world class medical centers), if you are positive you get sent home with no therapeutics and with instructions to go to the hospital if it gets bad. And in fact you are prohibited from getting Ivermectin which has shown an 84% early cure rate in cases diagnosed early. If I lived in a third world country at least I can get some of the drugs showing promise.

If I felt like I had it, I wouldn’t get tested unless it was a home test. Why throw yourself into the political morass of contact tracing and all the other state sponsored harassment. The best thing is be healthy to start with and make sure you have enough zinc and Vitamin D in your body and just stay home if you feel ill.

Carlo, Monte
Reply to  rbabcock
April 24, 2021 7:31 am

americasfrontlinedoctors.com will prescribe HCQ on-line.

Reply to  Carlo, Monte
April 24, 2021 2:18 pm

What about ivermectin?

Also, I’m curious if there may be pharmacies refusing to fill such prescriptions. That just occurred to me recently, that there might be.

Bruce Cobb
April 24, 2021 6:36 am

I don’t know how it will help patients, but each time one of these expensive devises gets used, the sound that the hospital hears is CHA-CHING!!!

April 24, 2021 9:27 am

They’re at it again!
The same Dr Evil labs in Oxford that became public enemy # 1 by making a low-cost covid19 vaccine hurting the profits of big pharma, have now added insult to injury by making a vaccine against malaria with efficacy above 75% for the first time.

The EU together with America’s FDA and Anthony Fauci are expected to join forces again to discredit the vaccine and prevent it reaching people suffering from malaria. The FDA dismiss malaria as being financially uninteresting. They have already ordered 10 million doses of the vaccine which they plan to sit on until they expire. EU officials expressed confidence in being able to find errors in the methodology of the phase 2 clinical trial of the malaria drug allowing them to discredit it worldwide. President Macron of France called it the “wrong vaccine made by the wrong people against a disease which is only imaginary hysteria anyway”.


Reply to  Hatter Eggburn
April 24, 2021 1:22 pm

Micron : “wrong vaccine made by the wrong people against a disease which is only imaginary hysteria anyway”.

Exactly what all this covid madness is.

Reply to  Hatter Eggburn
April 24, 2021 5:33 pm

The EU together with America’s FDA and Anthony Fauci are expected to join forces again to discredit the vaccine”: Who is the implicit subject of the passive “are expected”? In other words, who is expecting this–you? If so, take credit–or blame, whichever turns out–for your prediction.

April 24, 2021 10:25 am

So they have a new technology that can detect physiological nuances that they couldn’t detect before, and will now diagnose those abnormalities as covid even though they probably occurred in other illnesses (and still do), yet went undetected, but will now be defined as covid even with non-covid etiologies. Got it.

Reply to  icisil
April 24, 2021 2:06 pm

Asymptomatic SARS-Cov2 always struck me as a bit of an oxymoron. How can you possibly have a syndrome without having any symptoms when a syndrome is, by definition, a set of symptoms.

Reply to  MarkH
April 24, 2021 5:34 pm

Because SARS-Cov2 is a disease, not a syndrome?

Granum Salis
Reply to  mcswelll
April 24, 2021 8:27 pm

Technically, SARS-CoV-2 is a virus, not a disease.
The illness is CoVid-19

Reply to  mcswelll
April 24, 2021 10:27 pm

SARS literally stands for Severe Acute Respiratory Syndrome

Jim Whelan
April 24, 2021 3:02 pm

So far as I can tell this detects a symptom but in no way guarantees that the symptom is caused by a particular virus or disease.

Reply to  Jim Whelan
April 24, 2021 5:34 pm

Right; needs to be used in conjunction with other tests, as do many such tests.

Gary Pearse
April 24, 2021 6:53 pm

Months ago on WUWT I suggested it might be interesting to characterize the virus physically so that ‘geophysical’ instrumentation might have an approach to destroying the virus – ultrasound, high mag intensity (MRI), high gravitational field, etc.

I got the idea from an experiment 50 yrs ago when Paul’s Hospital in Vancouver kept a terminal peritonitus patient alive for many days in a hyperbaric chamber. It stopped bacterial action but other interventions they attempted during this hiatus were unsucessful.

Ideas like this are unlikely to come from biochemistry, microbiology and medical science (guy with hammer sees every problem as a nail).

April 24, 2021 6:55 pm

I’ve noticed, from personal experience, that ultrasound devices are being used for more mundane things like placing a large IV needle or catheter in a vein. A portable ultrasound device, the size of a cell phone, reduces the attempts to one and only one. Amazing tech.

%d bloggers like this:
Verified by MonsterInsights