Speckle Tracking in Pneumothorax

By Michael Prats

Speckle PTX Graphic

Speckle tracking quantification of lung sliding for the diagnosis of pneumothorax: a multicentric observational study

Intensive Care Med September 2019 - Pubmed Link


Take Home Points

1. In this retrospective and carefully selected population, speckle tracking identified the abolition of lung sliding with perfect accuracy.

2. More research is needed to see how this performs prospectively and with a broader range of operators.

3. This technology is not currently available for real-time clinical use due to constraints with current software and equipment.


Background

We know the ultrasound is great at finding pneumothorax. It has been shown to be better than xray for this diagnosis plenty of times. The problem is that maybe it is not as easy to perform for everyone, maybe the subjectivity makes it hard for some people to learn. Based on this idea, these authors hypothesized that a more quantitative measurement might be even more accurate. They had the idea to use speckle tracking on the pleural line to visualize and measure the movement of the of pleura sliding. Speckle tracking is a technology that recognizes unique tissue signature and can then track its movement over time, spitting out a graphical waveform of the tissue movement. It’s main use had been for cardiac ultrasound, including wall motion abnormalities. In any case, these authors speckle-ulated that this technology was going to enhance the identification of abnormal lung sliding in the evaluation of pneumothorax.

M-mode for Pneumothorax article we mention (by our very own Dr. Jacob Avila)


Check out The Evidence Atlas - Pulmonary for a review of prior evidence on point of care ultrasound in pulmonary diseases.


Questions

Can pleural strain values in normal lung be differentiated from those in patients with pneumothorax?

Does the measurement of speckle tracking show good inter-rater reliability?

Is speckle tracking more accurate than B-mode for determining lack of lung sliding?


Population

Two emergency departments, two intensive care units in three university hospitals


Inclusion:

  • Diagnosed pneumothorax

  • Lung US performed and images stored

  • CT scan performed

  • Unilateral pneumothorax


Exclusion:

  • Bilateral pneumothorax

  • No cardiac monitoring on US loops

  • Loops < 3 seconds

  • Images obtained by anyone besides the four investigators


Design

Retrospective analysis

Patients with suspected pneumothorax had received standard cardiac and lung ultrasound (see Scan below)

This lung ultrasound was compared to a CT of the chest that was done during this visit.


Main outcome was ROC curve to evaluate performance of the “maximal longitudinal pleural strain” to detect lung sliding abolition.

Therefore, they planned to first find the best cut off for optimal accuracy, then report the accuracy with that cut off.


Who did the ultrasounds?

Any of four study investigators could have done the scans.

There was only one fourth year resident who performed the speckle tracking. He had received training by performing speckle tracking analysis on clips of 5 normal lungs and 5 pneumothorax lungs.


The Scan

Linear probe with speckle tracking software

Linear

Protocol - single view bilaterally

  • Anterior inferior - 4th/5th intercostal at mid-clavicular line (while in supine position)


Learn how to do Echo for Pneumothorax from 5 Minute Sono!

5minsono


Check out Pulmonary Pathology on the POCUS Atlas!

thePOCUSAtlas


Results

N = 52 patients = 104 ultrasounds (2 sides per patient)

  • 80.8% of these were traumatic

  • Mean age 43

  • 39% were complete pneumothorax


Primary Outcome

Mean Maximal Longitudinal Pleural Strain

  • Abolished lung sliding - 1% (+/- 1)

  • Normal lung sliding - 46% (+/- 32)


Best cut-off was 4% which yielded:

Specificity 100% (CI 93-100)

Sensitivity 100% (CI 93--100)

+LR infinite

-LR 0

AUROC 1.00 (CI 1.00-1.00)


Other Findings

B-mode

  • Sensitivity 95% (CI 88-99)

  • Specificity 100% (CI 92-100)


No difference in the mechanically ventilated patients.


In normal lung sliding:

  • Intraobserver agreement 0.88

  • Interobserver agreement 0.18

In pneumothorax:

  • Intraobserver agreement 0.74

  • Interobserver agreement 0.52

No difference in intraobserver agreement between senior and junior physician.


It is stated that the speckle tracking was faster than B-mode, but I do not see that compared and would find it hard to believe. Based on the videos in the supplemental material (they are cool and you should check them out) - this takes a few minutes longer than just looking at the pleural for 1 or 2 breaths.


Limitations

Retrospective, small study. Need to test other populations, other (less skilled) operators. The population was 85% men and average weight of 70 kg. This may not be representative of your population.

It turns out that this can’t even be done in real time! According to authors, speckle tracking software (designed for echo) will not work if it does not recognize a ventricle or if the linear probe is being used.

This is great work for deriving the cut-off of 4%; however, in a way this was determined by just crunching the data to find the best fit. Now it will have to be prospectively analyzed.

Authors note they are not sure how well this would work with a lung pulse. Lung pulse is the pulsation of the heart seen in the pleural movement and also rules out pneumothorax.

They excluded more patients than they included. There were 139 eligible for the study, they excluded a total of 87 (and included the remaining 52). They excluded 51 due to the sonographer not being one of the study investigators and they excluded 32 because there was no cardiac monitoring - not sure why that would matter. Perhaps the speckle tracking had to differentiate lung movement from cardiac pulsations? The authors should have discussed why these were excluded. In any case, there could easily be some bias in this population and this represents a best case scenario since they took only ultrasound done by skilled operators.

Interobserver correlation was low in normal lung sliding (0.18). Authors say maybe this was because it is up to the operator to choose the respiratory cycle so maybe there is some variability. This would be a concern, especially if we attempt to extrapolate the quoted accuracy to a more heterogeneous group of operators.


Discussion

Is this fixing a problem that doesn’t exist? The authors cite that “In some situations, the diagnostic workup for pneumothorax remains challenging.” I’m not sure that is true. If you don’t see it on ultrasound, it's probably not very large and you wouldn’t do anything about it anyway. If you are still concerned, get a CT. What’s the problem? The authors cite a low sensitivity of B-mode (81%) allegedly due to the “qualitative” measurement. It turns out there is a good amount of variability in the sensitivity across different studies. This may be a technique issue. Low sensitivity usually means that people are missing pneumothorax when it is actually there (false negatives). This means that they interpret no lung sliding as lung sliding. I guess that speckle tracking would be a reasonable way to fix that problem, but so could more practice and training.


Take Home Points

1. In this retrospective and carefully selected population, speckle tracking identified the abolition of lung sliding with perfect accuracy.

2. More research is needed to see how this performs prospectively and with a broader range of operators.

3. This technology is not currently available for real-time clinical use due to constraints with current software and equipment.


Our score

3 Probes


Expert Reviewer for this Post

Soucy

Zachary Soucy, DO, FAAEM @ERDr_Sous

Zach is an Emergency Medicine physician and director of EUS and the EUS fellowship as well as co-chair of the system wide PoCUS committee at Dartmouth-Hitchcock Medical Center in Lebanon, New Hampshire. His areas of focus include ultrasound in medical education and system wide PoCUS.


Reviewer's Comments

What a great review guys! So far as the study goes, you guys covered all the strengths and weaknesses and perplexities I wanted to highlight. Ultrasound G.E.L Podcast bestowed upon me the great honor of reviewing the speckle tracking for ACS article a few months ago. Speckle tracking is such a cool technology that will undoubtedly find a greater number of applications. For those not familiar with speckle tracking, it is essentially a measurement of ultrasound movement in a multitude of directions. US image speckles are hyperechoic spots that have been likened to a fingerprint for each tissue segment. Each spot can be tracked in 2D or 3D depending on how much post processing you want to take on. When looking at hearts there are various directions each spot can be tracked to give a strain intensity. Personally, I am not doing this in my ED, and I know very very few who are because it requires a higher level of instrumentation/software, very good images, and time. That said, in experienced hands this technology can be tremendously informative.

In this study authors took a technology developed for heart movement and applied it to the lung. Due to the exquisite sensitivity for tracking spot movement they hypostatized when applied to the lung it could essentially give us a highly accurate yes/no answer regarding pneumothorax. As discussed above, there are many unanswered questions regarding selection bias, training, and whether someone could roll out software to allow real-time evaluation. That said, I think the author are on to something and their small trail does show admittedly limited but promising results.

As I discussed in my post last year, these “dream big” concepts and application of a variety of new and emerging ultrasound techniques to different tissues set the pace for innovation. What we look at as futuristic now may not be all that far in the future for the emerging “techy” generation of early career physicians, residents and medical students. Some of the largest companies in the world have entered the medical arena in recent years; think Microsoft, Amazon, Google, Apple! Their vision of medicine is, well…a bit different than most of us were taught in medical school.

My institution has embraced and is planning for many of these changes. I work teleED shifts a couple times a month where outlying smaller EDs call in for a host of services. Its really interesting work and a wonderful service for our rural area. Our Connected Care Center (CCC) hosts multiple specialty tele services including critical care, psychiatry, stroke, dermatology and on and on. Our CCC director notes that by some estimates, outpatient tele visits could outpace office visits by 2023! I also co-chair our system wide PoCUS committee as our home institution and affiliates find themselves talking the sono-lingo more and more often.

When I graduated medical school in 2008 I had not heard of any of these services and most of my current jobs did not exist. I believe it was assumed medicine would continue on its trajectory, similar to the prior 50 years, immune to disruptive innovation. Those days are clearly over and the US communities are a big part of medicine’s evolution. The recent alliance between Partners Healthcare and Sonosite as well as emergence of Butterfly iQ evidence the increased emphasis on artificial intelligence and usability of advanced US applications. As the number of private companies enter the medical space rises I expect these fringe studies to garner more attention and development.

Medicine may just be emerging from its chrysalis…I am interested to see what timid creature takes flight.


Cite this post as

Michael Prats. Speckle Tracking in Pneumothorax. Ultrasound G.E.L. Podcast Blog. Published on November 25, 2019. Accessed on October 01, 2020. Available at https://www.ultrasoundgel.org/81.
Published on 11/25/19 06:00 AM
comments (0)