PACS ADMIN SERIES • Tools & Tricks of the Trade • Image Quality & Monitor Calibration

Just another WordPress site

PACS ADMIN SERIES • Tools & Tricks of the Trade • Image Quality & Monitor Calibration

hello this is Herman oscillate and at the site of the EOTech exclaiming class in Dallas and the doula training class Nativity have a system like an altar or all of our guest speakers the talk about image quality and monitors and as many of you might not be able to make it their way to Dallas before being you ask him to demonstrate some of the highlights that you can take home me to him and really apply when you talk about monitors and image quality so first of all welcome Ken Thank You Pickett and Thomas but those are the first blood test petals in general what people should be using the test patterns in coming from the American Association of physicists medicine pattern and I recommend all the time is the QC patterns a starting point for all facilities it looks very similar to this empty pattern for those that haven’t must form a background in history but it has a number of attributes have been added to it that make or medically inclined to to find faults in the image quality itself what we did when we did this because I was a member of the task of a tena develop this whole process as we’ve made basically two types of test patterns for the two primary sizes of display what we’ve heard to them because CRTs were still in the industry about 1,000 line display in 2000 line test pattern okay one thing you have to always keep in mind with test patterns is they have to be shown at the exact 100 percent size if they were in format of that otherwise you lose the information that’s supposed to be derived from it so we have one this left-hand water here a very basic it’s the QC pattern for 1000 display and the 2 5 megapixels to the right have a 2000 – play test pattern okay so fundamentally whatever objects are on this are being basically twice that size on the 5 Meg and the reason for that is we have to double the pixels because the pixel size on the commercial like a 2 million pixel display or roughly 1/2 of it finally in pixel displace different pixel sizes do that so you don’t want to ever show a test pattern it’s something that’s reduced basically downsized with magnification or increased in size because you could make a good display look bad or a bad would look good by altering that so there’s a number of things in here because the the error of this particular test pattern was such that it was CRTs were still out there there’s a number of items within the QC pattern itself that are really not germane to flat panels because they just don’t suffer from focus or John problems here so things that are meant to be square will show up square out on any flat panel but there’s the basic keys on this and from the 17 people looking at this empty or prior remember a 95% square embedded within a 100% limits square the luminous 200% is really whatever the display is set to and then there’s a black square here with a 5% in here well what this should look like visually to your highest equal acuity other words this one in a bright block should be just as easy for your eye to pick up as the one at 5% and both of these squares are actually the same number of digital driving levels different from their background which turns out to be actually 13 Digital driving was out of 255 so if you think about that in terms of looking at typical medical display is that’s roughly equivalent to at least 26 what’s referred to as just most with different systems so they should be very visible to your Amish there should be no question about their their what are they and don’t see them like he is proponent because of the lighting the lighting and camera makers are because viewing distance for this would be roughly 2.6 meters for good doing what we did different from the senti is with these squares that go around this outside area or grayscale blocks that are increasing illuminance this is not life in this 100 sent this would starts it basic at the 8-bit level this one stops at 248 so there are eight differences between these or digital driving levels not your best behavior so you visually when you look at this you see a very distinct difference between each one of these squares which is all well and good and most commercial displays if you look at this one one 1000 line display two things of HD format you can see a difference between all the squares fairly good and there’s a nice difference between the 5 and the

95% what difference in terms of a test pattern from a medical display versus the commercial display is we added some small attributes within the test pattern itself that would mimic what it would take for your ID detects a long knowledge or a lesion that was very subtle and those are found as low contrast targets in each one of these very scared scale box and you can see this when the character on a hard time picking it up but there’s a square down there that is actually for digital driving levels lower than the the squares of value and opposite corners are the same and then they’re opposite corners here are plus for digital driving levels so that’s roughly 8 just muscle differences so these should be visible in these individual yes so you would each block will be in the quantumness feelings with those lawyers all the way around and when you have a much brighter display a lot of times your eye is adapting to an average luminous so sometimes they get a little darker once it’s extremely difficult to see the squares low contrast targets there’s down here generally speaking if you can see that majority these all the way around on the medical very display you’ve got an excellently calibrated display the luminance is very well distributed throughout the whole range of the display the thing that happens in a commercial displays which are not actually calibrated to weight medical displays are and that’s one of the key differences medical displays all follow a log linear function in terms of their health foot in response to the digital driving levels commercial displays in the other hand are very different they’re meant for office environments so a lot of times they’re very strong at the top and they’re very bright for a bright office environment they’re ready for it to low end so if there’s an calibrate cliana so for example this one of the key differences between the commercial and their medical displease because they’re not calculate this viewing in the dark area you would see not difference agentive my creativity what you tend to see is a very writing brighter area here the less difference is over here but right now I have the calibration on this one disabled and if I go in here and enable it you can see an immediate jump and change in the actual darkness of the image it becomes darker because we’re correcting for the lower and more than anything else all displays are sluggish so if you don’t necessarily to look at at Colonus there’s also this help of this text on the button right to the right what we’ve done here with this is a nice quick visual test for that I mean getting into the looking the corners that sharply first squares at zero level for its black and then this is at the mid-level one way of its 1:28 and this is of 2:55 for the background level and because this is the brightest square the L and the word control is one digital driving level download 255 and each subsequent letter going backwards is one more step lower and because this is the mid-range one brighter than this surrounding and generally speaking of where you well calibrated medical display should be able to see obviously all the word of quality if you can’t see that you have a real see at home generally speaking the agro elf control is visible here and because it’s brighter you may have a little bit closer the Oro L and control the mid mid range is fine the low in tends to be the hardest part that’s mainly because you have other bright areas around here and your light has adaptive upward pressure and just like any radiologist has done in the past with film if you’ve got a very bright spot to film they take out a piece of paper from their back pocket that’s black you talk about the bright areas so they can see the target rays so the first if you if you want to check quickly whether your monitors took me on a conviction or not you start up a can I see quality control those key blocks and then look at this brightness and it should give you a good indication – yeah if you go back four letters back to the T – t which would be for the children and the most difference that’s the same differential with the annoyance or wears in the corner that are plus or minus above below area so he can’t see this is questionable to hear you start looking in here in these squares and what you’ll find out a lot of times with displays they may be very well calibrated to implement and we have a very good distribution of luminance responses and they’ll be weak in the mid section you won’t be able to see them up here too well and then they may overemphasize it down here so you’re looking for balance and this the five and a 95 is wonderful for starting points for balance so if you think about display let’s say you’re given of your system and there’s a radiologist that would suggest at home you would recommend that they have that IV system that’s one of those we are checking for the tax in the corners fuel you know I actually elevation or not so it’s a visual check it’s purely a visual check there’s no light-years involved as ice your your ability to see with your own on human visual system a big details within the image yeah the vendors out there actually have this this test pattern build into their QC software so it’s like this is pulled up readout of the vendors of my Houston’s place there’s

one other attribute for this that’s still specific to flat panels investment again it would be used for see artis as well but they’re not use anymore and that’s a a grayscale ramp and that would be from the bright to the darkest value and the other side is the opposite from dark to the brightest and because this is of the mm line test pattern it’s starting out at 255 in here there’s four lines for each value as you step down through here on a 1000 line displays it’s two lines for each value of step through from 0 to 250 and what you’re looking for there is a very smooth transition in grayscale so you do not want to see what looks like a sharp edge factor like there’s a large jump and luminous difference it should look like a very nice smooth transition and if you see oh oh my it would look like a lineman to do is a large difference between two areas – that would be what would be look like there’s a contour line saying the upper part of the lung field so if you see contour lines of the lung field you don’t think it’s actually there’s on an anatomical feature that’s really present would be to throw this test pattern up and look at your face go away and just see how well that looks if it’s not smooth and that’s you’ve got a calibration problem now this is a commercial paper monitor this area medical date monitor and I know that you practice many hospitals might use commercially monitors for example the devil are what you are but how can you make sure that the Michigan wanted to accommodate how would you go about that well you could look at it with the basic test pattern with any kind of civil fewer you could look at the test patterns from the DA teen group okay it can be thrown up by use of a viewer that’s very simple you could read DICOM images and can read any other kind of JPEGs or tips and you can put the test pattern up and see how well it is calibrated because if i disable the calibration on this one you notice that difference in the luminance levels of course though the entire test pattern itself most of the word control down here is gone I can still see most of it here but in this case because it’s acting more like your commercial office product the very bright areas are watching animals and diary were natural and that’s very typical of them because you have 0 to 255 digital driving levels available a lot of times a commercial display will hit max illuminance at 200 to 230 digital driving levels and it’s just flags from that point out so that’s where you start losing information now let’s say go buy not to share should be monitored out any tools that Google recommended they’re used for coming yeah there’s some you can do it in a software-only approach or a combination of software hardware now this particular device I’m using is made by real vision it’s a black box airplane that does electronically what is more generally found within a medical display itself it’ll automatically does that so I’m taking a signal off a very standard commercial graphic card and running it through this little black box which talks to the computer and established a calibration for that you know based on the light meter feedback so I looked at the natural responses display compare it up to the DICOM grayscale response and you map the two things together to create a Limits response and then you would probably that’s this is a very simple white meter this basically sells for under $150 a lot of photography stores a dog food it’s very inexpensive but it’s quite accurate really so like me to the box you can make something to get close to commission or close or you can come very close what I’ve done this is in it’s enabled operating version and if I do a conformance test voluminous what about find is that the conformance curve for this display is actually within a plus or minus 10 percent tolerance which is considered diagnostic but you have to keep in mind because it is a commercial display it doesn’t have the luminance response to create a luminance ratio that’s appropriate for medical I’m sure where we’re working with 250 to 300 one rate ratios over medical displays a lot of times you can get up to 200 to one know any guidelines for what the solution wanted to you was well they have the FDA has some real strict rules in terms of mammography specifically whichever five million pixels and that’s because of the density of the information for a target for film source or our digital detectors when you’re looking at just general chest chest imaging musculoskeletal images a 3 megapixel display is quite appropriate it’s considered a 2,000 line display okay which would be the next step down from this particular devise excitability minor pictures but when you get into the ultrasound or the sliced images like mr or CT or you’re only working with an image it starts out at 256 square pixels you’re not working with a very high spatial content there’s to start with

now they do have a phenomenal dynamic range which doesn’t fit their Pro Minister to commercial displays however you can put newer sliced pictures up here images up here and through the digital tools of window leveling you can extract the information out and have a presentation even with the eight bits and just basically concentrate on the area of interest within the image itself for the dynamic lens show every sensor so you can do just about everything one of these I know there’s hospitals out there a chest film on 1 million or 2 million pixel displays but they’re considered clinical clinical level reads of ammonia in hospital another question that this is a sometimes a software issue and I understand sometimes there might be you before who cook tables or different processing that can be applied by defenders on the impact systems so when we want to make sure that we identify those issues as well we don’t necessarily probably at the monitor on the monitor better kinda go early in the pipeline right that’s true there’s been a number of instances that I’ve seen as in like in solar capacity where the various look-up tables that occurred in the entire pipeline like from the modality look-up tables say it’s a sewer scr’s device and then you have your packs software which has woke up tables worship your CR as the modality modality with anatomical tables that it picks are based on what part of the down when you’re looking and optimize the image and that has to be then piped in through the into the PAC system itself which has its own set of look-up tables occasionally for different things and then you get into the act lookup table which we refer to as the grayscale display function which is too hard the DICOM standard and the problem you run into a hospital facilities if say a patient was in issues with a cor and also an MRI and it was a one lesion in there and it showed up on him are either did show up on a CR machine you’ve got a question there which which one is right and why did one miss it the other one didn’t and it’s as I’ve seen where it was roughly the elephant in the room and it was missing on the CR device and it actually turned out to be a pax lookup table which needed a basically a software patch to put on it to fix the issue so what you end up doing is you have to isolate the actual workstation from the PAX you can do that first make sure the workstation is actually calibrated properly monitor to monitor actually up the test better right pull defender that does verify to the money time and then you won’t do it once you analyze this and you’re satisfied that the actual display function between your computer graphic card and the actual image file for the test pattern is correct it’s presented correctly you’ve pretty much eliminated that it’s a possible source of the problem at that point you can take the QC test pattern or one of the three on topical test patterns that are also supplied by a APM and inject that into the front end of your system as if it came off to CR or the MRI machine and then you just let the pipeline to treat it the way it normally treats it and see if the image that looks corrupted so and then you can do that because there’s a daikon representation with the SNP team of those images of beautiful and then become me we’re going to CD and they suited into the tax system probably about you know as a test system so it’s a testament and go for the whole they get up in a few one right we should see the same as if we would not use their software right because the the teaching 18q success pattern is in a dike on floor master on it both anything a 16-minute tip you can inject a test pattern that’s been in addition to an anatomical test penetrated test image okay into the system and see what it does to it and if there’s something wrong with a lot that it over corrects for some reason some area of the of the response away we’re not the darkest to the brightest areas you’ll notice the difference in the image and these tests that is the the publicly available to download from the internet but that’s correct the original chairman of the aap MTG 18 was at Duke University and the Duke University website now it’s where you can download all the test patterns and the document itself that explains everything okay so in the documents I mean even if you forget a little bit about all the characteristics we can even find a background how to interpret it and what there’s yes there’s appendices within the original document that describes every nuance of this test pattern how many pixels form a small square there and we have a plus or minus nominee or digital driving levels agree so sometimes this year questions on the whoops and maybe some people ask you that how long does this last so how would that find out when I would replace Jack the lifetime the monitor in terms of the luminous level which is

what you’re looking at for a specific set 24 calibration you have to look at what the original specifications for it and we as a as a hardware person from the high back we looked at the lamp sources and they rated them as 50,000 hours to their half-life so if you drove them then 100% of the roadie rated drive voltage and just let them deteriorate over time which is roughly 10% the first few thousand hours and it slows down quite a bit you can originally three to five years out of these displays easily it really depends on how you use them in your facility five years meaning penny for example yes it would be small we don’t generally do that because when the manners are calibrated by the factory they don’t drive 100% we always targeted 75 to 80% of the lamps actually under sent out money as a starting point so he said we always had plenty of room to grow underneath that to increase the drive to the lamps to bring the light output back though but it’s very important for a facility to remember to turn them off if they’re not used to turn them off don’t drive the lamps it’s not like your old CRT when you put a screen saver up there and a little icon would float around the screen and you said I’m really saving a life my CRT who actually work but with flat panels it’s very different story the lamps are always running at 1% of others are set to okay so what we do is grease say we’re running or not there’s still 11 whole string can be black and turn it off the lights are still on tuning off meaning not to sleep mode because we need to make sure that no power to the unit so they can still have power so it was that succinct signal come in and turn itself back oh you know which would be not stand by what your sleep mode for the Echelon lamps are turned off and there’s really no concern with the warmup time with them because all medical displays would be concerned with a commercial grade medical displays have a closed loop control system for lambs so the movie eternal one is constantly correcting for any error in the light output yeah and I think you told me before that if you wouldn’t do that that probably takes about a half an hour just leaves like this man look at 20 minutes for it to come close to anything stable and the test we ran it took almost hundred twenty minutes for the lambs to stabilize that plus or minus two percent the only other given joint that might be actually consolidation for commercial big monitors might make you sister mom or foundation it’s abundant of money and Mister commercial aids monitor that there’s not that the closed-loop shooked you might want to look anything diagnostic for the first 15 no you want new one definitely your own calibrate was either software solution or hardware so first looking what a little warm up a lot for at least 30 minutes we’ve achieved some kind of thermal stability in the back light okay so kindly mention we have this commission big monitor every what this is like inter and that moments is involved what’s involved it did come begging hardest to do pain she was a little bit sure we’re not going to actually calibrate this we’re going to a conformance test it okay approve what its response is I’ve already actually calibrated this one when you working with a commercial grade display so if our packages will generally default some number of test points just start out with that fifty-five available normally speaking is somewhere around 17 or 18 s what they start with they depending on the quality of the display and I did a white paper for the company has this control box it’s actually controlling this one I found that that particular commercial display would not calibrate properly or even close at 32 test point 64 or on or 2410 I had to go to all 255 together a somewhat accurate response in contrast to that this particular brand here which is higher quality display I can do this one with with 32 points so I can get an excellent response curve out of it again that’s just a difference in the quality so you have to be careful you buy it by the cheese you want to buy it yeah all right good all the way to 1 all over my agenda okay so what what it’s already preset the ghost is a it clicked I need to have the light here basically held to the center of the screen no just click OK here and it’ll prepare the test you would have to start doing you don’t block the screen out the actual software I’ll take you to a flea control the display oh and I’ll start with the darkest level and in this case it’s only going to do the incremental we test in large steps okay so you must see large steps so this is kind of course performance test but it gives you a good idea what it looks like okay so this you can do this for all 255 steps but you’d be sitting here holding this thing up for a few minutes one thing to keep in mind any time you’re working with the blue crystal display solution for the light Peter is never add extra pressure to the screen you know it was going to put pressure on it you squeeze the liquid crystal material out of the way and it gets brighter and you’ll distort your results dramatically so this to go mean I committed this oh yeah that’s good that wasn’t that many test points so I go back and I want to go back to reports latest the ceiling function but controlled by

that performance and there’s my test results okay and it’s set right now with a customized 10% band which in cameraman up they’ll pick this up in terms of parallel lines basically sweep down these squares that are representative of their target values we want to hit and the exits are would have actually measured so we have two points basically they’re great on the line but again this is the diagnostic tolerance line for customized 10% if I changed that to secondary tolerance plus minus 20% for clinical referral type applications all the data points for this display are well within that parameter so it in that that sense of the term being responsive to the grayscale display function it’s accurate there’s actually not a lot of work to do this is you know you look at the the QC test panel and if needed into one of two minutes about doing the air calibration that’s not we’re assuming that that definitely is feasible or whatever they’re not stupid so he’s not a doctor who do this it is almost very little instruction okay thanks well thank you very much Ken