Kuro5hin.org: technology and culture, from the trenches
create account | help/FAQ | contact | links | search | IRC | site news
[ Everything | Diaries | Technology | Science | Culture | Politics | Media | News | Internet | Op-Ed | Fiction | Meta | MLP ]
We need your support: buy an ad | premium membership

[P]
Why JPEGs can be bad for your Health

By porkchop_d_clown in News
Tue Jun 20, 2000 at 07:20:21 PM EST
Tags: Science (all tags)
Science

According to the The Economist (use cypherpunk/cypherpunk to login), the European Heart Journal and the Journal of the American College of Cardiology have published a study on the impact of using the JPEG image format to transmit diagnostic images.


The conclusion is that using JPEG compression ratios of more than 6:1 increases the risk of misdiagnosis. At 16:1 the error rate in detecting disease in an angiogram increased by 30% in one study and, in another study, 45%!

The problem is caused by the nature of the JPEG algorithm. JPEG is intended for the transmission of photographs and achieves it's high rates of compression by (in effect) smoothing the data. The assumption is that the loss of fine detail is an acceptable trade off for increased transmission speed - particularly since the smoothing operation hides the fact that detail has been lost.

Unfortunately, discarding fine details - and hiding the fact that they were discarded - can be bad news for medical imaging.

I've looked for online references to the studies themselves but I haven't found any. Can anyone else track them down?

Sponsors

Voxel dot net
o Managed Hosting
o VoxCAST Content Delivery
o Raw Infrastructure

Login

Related Links
o The Economist
o Also by porkchop_d_clown


Display: Sort:
Why JPEGs can be bad for your Health | 29 comments (25 topical, 4 editorial, 0 hidden)
Disturbing... (4.50 / 2) (#1)
by fluffy grue on Tue Jun 20, 2000 at 04:35:35 PM EST

It disturbs me greatly that medical professionals could even conceive of using, much less trusting, a lossy format for transmitting medical data. I mean, okay, so they're not clued-in to the technical details, but don't they NOTICE when an image starts looking blocky before they send it?

We need pervasive, widespread use of PNG, it seems.
--
"Is not a quine" is not a quine.
I have a master's degree in science!

[ Hug Your Trikuare ]

Re: Disturbing... (2.00 / 1) (#4)
by porkchop_d_clown on Tue Jun 20, 2000 at 04:39:54 PM EST

Apparently, the problem is that details are lost long before the file is so compressed that it develops the "chunky" look. It's like airbrushing out the centerfold's freckles. Good for centerfolds, bad for angiograms. *grin*



People who think "clown" is an insult have never met any.
[ Parent ]
Re: Disturbing... (none / 0) (#5)
by fluffy grue on Tue Jun 20, 2000 at 04:46:34 PM EST

I guess that makes sense, but still... why does medical imaging software even give the OPTION to save as JPEG? Even if the radiologists et al don't know any better, the programmers of the software should. (Then again, in mye xperience, most such specialized software is written by idiots, sadly.)
--
"Is not a quine" is not a quine.
I have a master's degree in science!

[ Hug Your Trikuare ]
[ Parent ]

Re: Disturbing... (none / 0) (#11)
by porkchop_d_clown on Tue Jun 20, 2000 at 05:56:22 PM EST

(Then again, in mye xperience, most such specialized software is written by idiots, sadly.)

Unfortunately, that's probably the key point. Although even if the programmers were good, they may not have had the medical training to realize the implications of using JPEG. Heck, I've looked at my own X-Rays lots of times. And they still look like ghostly ink blots. I know I'd never notice that compression was degrading their fine detail.

And, you forget the impact of management on software projects:

Boss: "Hey, the radiologists are complaining that it takes to long to download the images!"

Programmer: "... But ..."

Boss: "No buts. The customer is always right!"



People who think "clown" is an insult have never met any.
[ Parent ]
Re: Disturbing... (none / 0) (#28)
by el_chicano on Sat Jun 24, 2000 at 12:44:45 PM EST

Boss: "No buts. The customer is always right!"

Obviously the boss isn't Bill Gates!!! :->


[ Parent ]
Re: Disturbing... (none / 0) (#20)
by Boojum on Tue Jun 20, 2000 at 11:23:25 PM EST

(Then again, in mye xperience, most such specialized software is written by idiots, sadly.)

Hey! I take offense to that! Seriously, I've writting some fairly heavy duty scientific imaging software for years for my father to use for his research. The images are photographed from the electron microscope and then scanned in with a fairly high-end slide scanner at high resolution. And they're almost invariably saved in uncompressed TIF form, (that being the form prefered by my software.) In general, the clarity is remarkable. On the other hand, my dad goes so far as to do color callibration and plot his own optical density curves, so perhaps he's the exception. But I'd just like to point out that not everyone in the medical community is incompetent when it comes to imaging.



[ Parent ]
With PNG, you could probably compress 2 to 1... (none / 0) (#21)
by pin0cchio on Tue Jun 20, 2000 at 11:31:27 PM EST

...and not lose anything, thanks to PNG's "Paeth" predictive filter and "Deflation" lossless coding.
lj65
[ Parent ]
yes but why? (none / 0) (#27)
by Anonymous 242 on Thu Jun 22, 2000 at 11:59:42 AM EST

You'd think that any system that works with multiple large images would keept the bulk of them in some sort of compressed archive such as .tgz. I also know that many database products have built in compression for certain types of large binary data files.



[ Parent ]
jpeg compression (none / 0) (#6)
by jezmund on Tue Jun 20, 2000 at 04:48:43 PM EST

hmmmm.....compression causes data loss? It's an interesting story I guess, but it rates a little high on the "obviously" scale.

Similar article on the same topic (4.00 / 1) (#8)
by madams on Tue Jun 20, 2000 at 05:08:18 PM EST

I read about this a few weeks ago on UniSci in an article called Are compressed images dangerous for medical diagnosis?.

Oh, and no registration is required to read UniScie. :-)

--
Mark Adams
"But pay no attention to anonymous charges, for they are a bad precedent and are not worthy of our age." - Trajan's reply to Pliny the Younger, 112 A.D.

Well, duh! (4.80 / 5) (#9)
by the Epopt on Tue Jun 20, 2000 at 05:14:17 PM EST

What part of "lossy" don't you see? If you want fine detail, use TIFF.
-- 
Most people who need to be shot need to be shot soon and a lot.
Very few people need to be shot later or just a little.

K5_Arguing_HOWTO
Re: Well, duh! (3.00 / 1) (#12)
by bmetzler on Tue Jun 20, 2000 at 07:01:56 PM EST

What part of "lossy" don't you see? If you want fine detail, use TIFF.

Exactly. What's the point here? Will the next story be, "Study finds that screw drivers are 70% more effective in driving screws then hammers"?

This is silly. I'm not going to quit using jpeg's to save images of my cat, just because they might cause a mis-diagnoses of some old grannies heart tests.

Was there a point to this story? Or have I just hit the end of my patience for "stupid" news today? Are doctors actually using jpeg's, and if they are, who will posting the story here help? Am I supposed to ask my doctor before each test whether the images are saved in jpeg format?

-Brent
www.bmetzler.org - it's not just a personal weblog, it's so much more.
[ Parent ]
Re: Well, duh! (3.00 / 1) (#15)
by Potsy on Tue Jun 20, 2000 at 08:39:13 PM EST

The news here is not that JPEGs are lossy -- everyone (at least, everyone on K5) knows that. The news is the way in which they are being used.

I never would have guessed that medical imaging used lossy image formats. I would have thought that people who put together such systems would have been smarter than that. I actually find this very disturbing, and I'm glad I heard about it.

I think the problem with this story is the headline. It should read something more like this:

"Stupid decisions by medical imaging companies could cause misdiagnosis"

Like I said, I never would have guessed this was happening. I find it scary, and I hope enough attention is paid to this problem so that it gets corrected. That's the point here.

[ Parent ]

The medical industry *requires* that images used f (4.00 / 1) (#19)
by your_desired_username on Tue Jun 20, 2000 at 10:03:57 PM EST

From the abstract of the <a href=&quot;http://www.harcourt-international.com/journals/euhj/previous.cfm?art=euhj.1999.2101&quot;> paper in question: 'Background The Digital Imaging and Communications in Medicine (DICOM) digital exchange standard for angiocardiography prescribes that images must be stored loss free, thereby limiting JPEG compression to a maximum ratio of 2:1. '

[ Parent ]
Re: Well, duh! (none / 0) (#13)
by feline on Tue Jun 20, 2000 at 07:02:35 PM EST

exactly, the only argument against using tiffs are bandwidth and storage space limits, and any hostpital that has a cardiology dept. is going to have a fast network or if they're transferring between hostpitals, probably a pretty snappy outside net connection.
------------------------------------------

'Hello sir, you don't look like someone who satisfies his wife.'
[ Parent ]

PNGs are lossless.. (none / 0) (#14)
by Inoshiro on Tue Jun 20, 2000 at 07:08:48 PM EST

And they compress better than the tiff format :-)

--
[ イノシロ ]
[ Parent ]
Re: Well, duh! (none / 0) (#16)
by Zagadka on Tue Jun 20, 2000 at 09:15:43 PM EST

Actually, some versions of TIFF supports numerous compression systems including JPEG. The file format that we normally call "JPEG" is actually named JFIF, and is only one of several file formats that use JPEG compression. My point is just that you shouldn't assume an image is lossless because it's a TIFF.

Anyway, PNG is a simpler format, and I wouldn't be surprised if it had better compression than typical lossless TIFF's.

[ Parent ]
Correct: PNG packs real tight. (3.50 / 2) (#22)
by pin0cchio on Tue Jun 20, 2000 at 11:36:59 PM EST

The PNG process in a nutshell:
  1. Run a lossless predictive filter called Paeth on grayscale or truecolor data to help take advantage of two-dimensional continuity.
  2. Deflate the result (this is essentially gzip).
  3. Wrap the PNG chunk structure around the compressed data.
Voila, 2:1 or better lossless compression on most images.
lj65
[ Parent ]
Re: Correct: PNG packs real tight. (none / 0) (#24)
by Anonymous Hero on Wed Jun 21, 2000 at 10:30:03 AM EST

I thought PNG just separated the red, green and blue channels and compressed them either separately or in a long stream of data.

Or maybe that was PCX. I forget.



[ Parent ]
old news (4.50 / 2) (#17)
by mercenary on Tue Jun 20, 2000 at 09:30:34 PM EST

Doctors, especially cardiologists, have been doing JPEG/MPEG diagnostic studies for years and years. They usually find that beyond a certain compression level, diagnosticians can't tell the difference.

Since we can't tell what kind of programs will evolve in the future that could utilize detail invisible to the human eye, we should definitely be wary of tossing out pixel information in favor of space or easy display.

Many of these studies, however, aren't about storing images as JPEG/MPEG but rendering them as JPEG/MPEG for the purpose of remote diagnosis.

If you want to find more, search PubMed for JPEG, MPEG, and Heart.

Also check out some older publications from JD Thomas:

Karson TH, Zepp RC, Chandra S, Morehead A, Thomas JD.
Digital storage of echocardiograms offers superior image quality to analog storage, even with 20:1 digital compression: results of the Digital Echo Record Access Study. J Am Soc Echocardiogr. 1996 Nov-Dec;9(6):769-78. PMID: 8943436; UI: 97098765

Karson TH, Chandra S, Morehead AJ, Stewart WJ, Nissen SE, Thomas JD.
JPEG compression of digital echocardiographic images: impact on image quality. J Am Soc Echocardiogr. 1995 May-Jun;8(3):306-18. PMID: 7640024; UI: 95367274


Article Links (2.50 / 2) (#18)
by Anonymous Hero on Tue Jun 20, 2000 at 09:46:36 PM EST

For a bit more info, you can read the abstract of this article here on the European Heart Journal website.

You can get the full text (if your organization subscribes to the European Heart Journal) here .

Resurrected Man (none / 0) (#23)
by Anonymous Hero on Wed Jun 21, 2000 at 06:37:10 AM EST

I vaguely recall a sci fi story about matter transmission called the "Resurrected Man"... i forget the author... Sean Peters?

Anyway the way people got about in this world was matter transmission booths in which people were taken apart, transmitted over a network, and reassembled at the other end.

Now this means transmitting rather a lot of data. So they compressed it.

Lossily.

And people occasionally got reassembled with compression artifacts.

just your disturbing idea for the day,
Francisco

Re: Resurrected Man (none / 0) (#25)
by Anonymous Hero on Wed Jun 21, 2000 at 11:38:50 AM EST

The name of the novel you're thinking of was "The Resurrectionist".

[ Parent ]
Well, DUH! (none / 0) (#26)
by jonr on Wed Jun 21, 2000 at 11:51:53 AM EST

It's the first thing that popped into my mind. More lossy compression == Less accuracy. Hello? As if the Doctors and the devlopers working on this system didn't know? Does the economist have to spread some FUD?
Sheesh!
J.

Re: Well, DUH! (none / 0) (#29)
by jovlinger on Wed Jun 28, 2000 at 02:36:36 PM EST

well, the 1:6 ratio was interesting.

IIRC, the JPG format doesn't mandate that the entire image be compressed at the same ratio. So it should be feasible for a doctor to highlight a region of interest, compress the whole thing (with the ROI not more than 6:1) and then email it to a collegue for a 2nd opinion. Johan

[ Parent ]

Why JPEGs can be bad for your Health | 29 comments (25 topical, 4 editorial, 0 hidden)
Display: Sort:

kuro5hin.org

[XML]
All trademarks and copyrights on this page are owned by their respective companies. The Rest 2000 - Present Kuro5hin.org Inc.
See our legalese page for copyright policies. Please also read our Privacy Policy.
Kuro5hin.org is powered by Free Software, including Apache, Perl, and Linux, The Scoop Engine that runs this site is freely available, under the terms of the GPL.
Need some help? Email help@kuro5hin.org.
My heart's the long stairs.

Powered by Scoop create account | help/FAQ | mission | links | search | IRC | YOU choose the stories!