PEPconnect

Acceleration Techniques in MSK MRI Webinar

The herein illustrated statements made by Siemens’ customers and physicians are based on their own and discrete opinion. The speaker is responsible for obtaining permission to use any previously published figures or tables. The speaker is also responsible for obtaining permission to reproduce any photograph showing recognizable persons.
The statements by Siemens’ customers described herein are based on results that were achieved in the customer's unique setting. Since there is no "typical" setting and many variables exist there can be no guarantee that other customers will achieve the same results.
Some products/features (here mentioned) are not necessarily commercially available in all countries. Due to regulatory reasons their availability cannot be guaranteed. Please contact your local Siemens organization for further details.

Hello and welcome to another applied radiology expert forum webcast. Today's program is entitled acceleration techniques in musculoskeletal MRI, improving efficiency, improving patient experience. I'd like to take a moment to introduce Doctor Jan Fritz, who is an assistant professor of radiology and radiological Sciences at the Johns Hopkins University and the Job Johns Hopkins Hospitals. Google discuss new acceleration techniques in MSK, MRI imaging, and share his clinical experience with multiple techniques. Doctor Fritz is a board certified and fellowship trained academic, musculoskeletal radiologist. With over 15 years of experience in interpretation and research of musculoskeletal MRI, he has authored over 150 peer reviewed Journal articles and book chapters. His research efforts at Johns Hopkins focuses on the development of clinical development and clinical integration of novel and rapid musculoskeletal MRI techniques, including two D and 3D MRI metal artifact reduction MRI. MRI, neurography and interventional Mr imaging techniques were very pleased to have him join us. Today, Doctor Fritz, do we have you on the line? Yes, I'm here, Karen. Thank you for having me. Lovely, thanks again for joining us. I'm curing Anderson group publisher here to applied radiology and I will be your host and moderator today. Today's program has been sponsored by Siemens Healthineers and we would like to thank them for their continued support of events such as this. Following Doctor Fritz's presentation. We will have time for a brief Q&A and we certainly encourage you to submit questions that you may have by the QA area at anytime during the event. So without further do let me turn it over to Doctor Fritz who will begin his talk. Now, Doctor Fritz, the floor is yours. Thank you very much. Thanks again for having me so in the next 45 minutes or so. Will discuss. The acceleration techniques that are currently available in muscular schedule. The MRI, an also the ones that are on the verge of becoming available or the ones that are that will be available in the near future. Here my disclosures. We at Johns Hopkins receive institutional research support from multiple industrial partners. So the objectives for this talk will be to apply acceleration techniques to create rapid clinical MRI protocols of the musculoskeletal system. Also, implement new acceleration techniques into already existing clinical practice protocols and illustrate a variety of post processing techniques applied to rapidly acquired MRI data, specifically in the 3D realm there. So let's talk about efficiency and added value to MRI and how this is happening. So what we would like to do when we accelerate our imaging protocols? We would like to retain and improve image quality by decreasing acquisition time. So we would like to also retain the number in the types of sequences that we're used to. So it's like a win win win situation that we're trying to create, which is obviously difficult. What this will do is we can increase the availability of MRI for example in countries or health care systems. Where there's a long wait time, we can create more MRI slots. We can also make MRI more tolerable if we decrease the time, we can decrease motion artifacts. We could decrease the need for anesthesia, for example in the pediatric population. We can also compensate for decreasing revenue that is currently happening, so there's a lot of reasons why we would like to increase the efficiency of MRI and it adds values of multiple layers. Now when we talk about exploration techniques. Field strength is an important consideration, while all the acceleration techniques are applicable for 1.5 and three TA lot work better at three T and that is not only because we have more signal in 3T, which we all know depending on the system in the course we use, but the acceleration is intrinsically better for 3T when we compare 1.5 and three T, we can image 4 four times faster already from the get go. So a lot of things work better with three T, But I want to emphasize it also works at 1.5 Tesla. Sometimes actually remarkably well. Now, scanner technology matters and a lot of the acceleration techniques are often better available on digital scanners rather than analog or older scanners. So it depends on what scanner you have available at home or in your institution, and sometimes new acceleration techniques need a certain software level to be to be running. Now, as with everything in this case, surface coils are very important. And if available, it almost always as a general rule, works better with surface coils. Now what those could be? Dedicated surface coils that are anatomically shaped like for the ankle and the foot for the need for the wrist, for the shoulder. Or it could be flexible coil that can be wrapped around the body part that we image, but the key is to have multi channel coil to meaning there's more multiple elements in there and also to minimize the air gap between the surface of the skin and the coil. So the first half of the talk will be about a 2D acceleration techniques, and then we'll turn to 3D acceleration techniques, and then we'll review some post processing techniques now inherit difference between 2D acquisition and 3D acquisition. Is that in the 2D acquisition we're acquiring Slice by Slice, so we do one slice and then we do a second slides into third slides. We may be able to interleave them, but we're basically requiring every image separately and then put them together, whereas in the 3D. Technique we always acquiring a volume for exciting a volume and then we get the signal out of that and then later we slice the data. So how fast can we go? So in the beginning of MRI and depending on the MRI scanners today we use unaccelerated MRI. That means every line in case main with sample separately and there was no skipping in the acceleration. And then what first came into play was parallel imaging. Parallel imaging means for example that we can skip facing coding steps on the left hand side here where it says Unaccelerated. It means that there's a frequency in a phase encoding. If you were to acquire every line here, you would have an unaccelerated image. If you were to skip every second phase encoding a line, then you would have a two fold acceleration or parallel imaging technique. Acceleration of a factor of two and the factor of two will cut your imaging time in half. Now scan time is directly proportional to the number of phase encoding steps, and so if you cut your phase encoding steps in half, for example with an acceleration factor of two or crop of two, this is 1 technique. Then you can do scan time in half. It comes at the penalty of signal, so you lose signal. But for example, if your MRI protocol has an abundance of signal or you scan it 3T, then often we can afford the parallel imaging twofold acceleration without losing image information in the image is still look acceptable or not even noisy. Here is an example of a twofold acceleration with grappa, which is 1 parallel imaging technique that was described in 2002 by a doctor Griswold. On the left hand side is an unaccelerated image on the right hand side is a twofold accelerated image which cuts the acquisition time in half. Those are T1 weighted images. Corona, Lee of the knee and you can see that the image information is rather retained their and the level of noise is almost not perceptibly different, so this is a good case for applying the crapper to acceleration and cutting your imaging time in half. Now this was available for a long time, and when you were increasing the acceleration to a factor of three or four, you get acceleration artifact, and. This was linked to me by Doctor Eck from and why you? And it shows that if you would accelerate instead of instead of a factor of two, you would accelerate to a factor of 4. You get the image on the left, so there is aliasing there, and there's an increased amount of noise in that image. Now you probably have heard about deep learning and machine learning and artificial intelligence and what the group in NYU does. They're taking these degraded images on the left, which suffer from acceleration artifacts and their applying machine learning. Reconstruction algorithms and they can with those new operation acceleration algorithms that can create images that look rather with high image quality despite their acquisition of crap out four. So machine learning will probably add another step in the acceleration, not so much that we are not able to do it. We are already able to accelerate fourfold with parallel imaging, but the images are too noisy with our current reconstruction algorithm. So if we're adding the machining. The machine learned machine learning algorithms that learn by example and that have a previous knowledge on how images of the knee should look like. We may be able to use for full acceleration with parallel imaging only and reconstruct high image quality like shown here on the right hand side. This is the same image one is reconstructed with the current reconstruction algorithms and on the right side is with a machine learning based reconstruction algorithm and the image quality is quite strikingly different here. So if we fill in our table, we have a twofold acceleration and if we use use parallel imaging, we can already cut this down in half. So most of our a protocols here, Johns Hopkins. If we use 2D accelerate to the imaging techniques have a two fold acceleration embedded and that is often economic and efficient to do now how to further accelerate? We just said we can do factor 3 infected for acceleration with parallel imaging, but with the current reconstruction techniques the images suffer from acceleration artifact so. Currently, without deep learning algorithms which are in the makings but to my knowledge, not publicly available yet, we have to do something else, so one way of doing this would be thinking about how we actually acquire faster to pursue an actual images. So we just said when we introduce the two DTS E or the 2D acquisitions that we do this slice by slice, we excite one slice and then we get one image and then we excite a second slide and then we get another image and we do this for the entire stack until we have our need together. Like in this example, so this is called successive image acquisition and it has been proven for many years. It delivers high image quality, but it is nowadays time consuming to do that way. So one way to do this faster would be to not successively accelerate acquire, but simultaneously excite the images. So you could send into RF pulses or two are frequencies and excite this slice on the that is the left. Yellow line and the other slides that is the right yet yellow line at the same time. If you were to do that, and it's technically possible then you would get 2 images. The problem is that they are over superimposed upon each other and it will be difficult to read. However, new algorithms like the Kuyper, any algorithm here actually allows you to put these images apart and then later calculate the two images separately. So what you have gained now is that you can acquire 2 images at the same time. For basically a factor of two acceleration. So if you had 20 images in your choir, them as packs of two, you would only have to used. You only have to do this 10 times instead of 20 times and you save another 50%. So you could also do three images at the same time or 4 images at the same time. Now we said parallel imaging the penalty you pay is signal here for simultaneous multi slides or multi Band acquisition. The penalty you pay is energy, so you have to excite several slides at the same time and it costs you RF energy and then at some point you reach your sore limits that are imposed by the regulatory. Societies and institutions. But if you stay within that, you can excite several slices at once now and we did a study on that and this image here shows pet means parallel imaging technique and SMS means simultaneous multi slice and we combine parallel imaging with simultaneous multislice acquisitions. That's especially helpful in MSK where you have to retain your contrast so when you do SMS acceleration you do it at the expense of TR and at some point you no longer want to reduce your TR your petition time. Because you would lose your your PD or T2 contrast. So at some point when you have accelerated with SMS enough, you want to accelerate further with parallel imaging because that is TR neutral in that sense. So here you can see at the at the bottom this is a path to SMS to acceleration, meaning there is a parallel imaging acquisition, two factor and the simultaneous multislice factor 2 in their combined 2 * 2 is 4. So this is a fourfold accelerated image with TD. 2D that's been interpreted echo technique here on the right hand side is a pad to SMS three. That means we we acquired three slices at the same time. Each of them with parallel imaging factor 2 acceleration, so that makes it a 6 and is basically shows that this is feasible. This was acquired at three T in level one, so it stays within the required star limits. Which shows that a sixfold acceleration is possible with this combination of that technique. That's quite a step compared to the twofold acceleration that we currently have on our scanners. We said before that there's there's a signal penalty if you accelerate with parallel imaging, whereas with simultaneous multislice there's a sort of penalty. Auras are limited, but very, very little signal decay, and this. This graph shows that. So this is the signal of the bone mirror here, of the femoral condyle, and with simultaneous Multislice acquisition you can see that the signal to noise ratios are higher when compared. On the left hand side to the parallel Imaging acquisition Factor 2 and Factor 3. Now if you combine two factor of parallel imaging and two factor simultaneous Multislice acquisition, you can see that your simultaneous Multislice acquisition is almost neutral when you compare it to pad 2 and pad 3. And so this is, this is approved by the images don't look terribly noisy. They actually look like normal images because you look through the simultaneous Multislice acquisition. You afford almost a signal neutral acceleration there. Now, here's some clinical examples on the top row is a PD weighted image, and you can see a small free edge meniscus tear where the errors is an in the lower row. There is T2 fat sat images which showed the fluid right and you can see that there is a partial thickness cartilage fissure there where the error is now the image quality stays relatively the same, but you can see on top for the PD or the intermediate weighted images that with the acceleration they dropped from 3 minutes and 10 seconds to one minute and 26 minutes seconds. Or the T2 fat sat drops from 2 minutes and 40 seconds to one minute and 15 seconds depending on what acceleration and parameter modification you use there. So this is quite an acceleration for clinical practice that's available now, so with that technique you could form and this is what we did here. A5 sequence 5 minute 2D PSC or FC MRI protocol and we see this here. This is a combination that works so the entire 5 sequences can be acquired in less than 5 minutes here and so this is a. Sagittal, PD weighted image. Sagittal T2 fat sat. Image A Corona LDPE defense at image coronal T1. Weighted image and PD axial image here and so every sequence is about one minute. This was acquired at three T and you can see the image quality appears like at least what we hear. Johns Hopkins would consider our acceptable daily high resolution image quality. Now here's some examples so I said before. This works at 3:10 at 1.5. This is an example of 1.5 Tesla 2D acceleration. This is a 5 minute new protocol. At the bottom and then at the top is a 10 minute version with copper to acceleration as comparison and you can see here that there is an enter cruciate ligament tear that is visible quite well on either protocol. So even at 1.5 Tesla you can realize a five minute 5 sequenced over spinecor fast spin, echo knee MRI protocol and I think that's quite the advancement so we can. Accelerate this to 5 minutes in clinical practice. Here's the 3T version. Different patient and there is a medial meniscus tear here where the yellow errors are and you can see this the same combination PDN T2 fat, sat, set satchels and then XLT defects at T1 and P defense at caroles. Here's another example of a medium meniscus tear with the tibial bone married Emma Anna Prepatellar Bursitis here also comparing the 10 minute parallel imaging protocol with the five minute parallel imaging SMS protocol. And the findings are very similar, and our radiologists in the study evaluation were able to make the diagnosis similarly well on both protocols. Here's a partial thickness tear of the Patella at the origin where the arrow points is also knows, as known as a jumper's knee, and you can see here there's a partial tear with the fluid is tracking into the tendon origin here. Here's another case over another. Enter cruciate ligament tear at 1.5 Tesla. Also acquired with a 5 minute protocol at the bottom and the 10 minute protocol up top. Here's another medial meniscus tear. Also comparing one point at 1.5, the 10 minute protocol in the five minute SMS. A parallel imaging protocol showing the findings equally well. Now our colleagues add an then why you combine the simultaneous multislice acceleration with compressed sensing, which is another way of accelerating. Its a nun dissembling technique that later recovers missing data with a smart algorithm and they showed that they can get similar accelerations with a combination of compressed sensing and simultaneous multislice acceleration as compared to parallel imaging and simultaneous multislice acceleration. So the next step would be to see if parallel imaging acceleration simultaneous multislice acceleration. Encompass sensing acceleration can be combined to even achieve further acceleration. So the full full acceleration is possible with the combination of parallel imaging and simultaneous multislice or multiband acceleration an from an unaccelerated 100%. You can drop your imaging time down to 25%. So if you're imaging the knee and your protocol was 20 minutes with the application of a fourfold acceleration, you could drop it down to five minutes and retaining your imaging parameters. Now we just spoke about cover sensing acceleration. And I wanted to quickly review what that is so. There's different ways of doing it, but one way of doing it would be that you sample your case based fully, but then in the periphery of your case face you under sample, and then you apply a smart algorithm and that could be a fixed algorithm or an algorithm that is based on machine learning that get smarter overtime and that reconstructs the missing information. So it's almost a little bit like free lunch, so you get to sample much faster. You get the image must faster, but your image reconstruction is a little bit more time consuming depending on the computer you have and the computer will reconstruct the missing information there. So you scan them patient faster and you reconstruct a little bit longer, so this is a paper that we wrote evaluate if that come presenting acceleration applied to see Mac, which is a metal metal reduction algorithm, is really the same, and so here. Up top is the conventional conventional see Mac algorithm and at the bottom of the conversation he make algorithm and you can see that they are indeed very similar. There's really surprisingly no visible or perceptible difference here, or is very minute. And so here's the combination of comparison of what is available, so these are sagittal images over patient with a knee replacement. And if you know this from practice imaging metal implants is rather challenging, so you need a technique like see Megan Maverick that are advanced to reduce the metal artifact and in the center column. Here is the is the regular parallel imaging accelerated. See Mac acquisition and you can see at this high quality level this has a high spatial resolution and it uses 19 phase encoding steps, so 19 phase encoding steps is a lot, but. You have excellent metal reduction here and it takes you 11 minutes per sequence if you apply the compress sensing algorithm then you can collapse the imaging time to four minutes and 28 seconds, which you probably would agree is much more clinically feasible than 11 minutes for one sequence, and so at our practice, the four 4 minutes is very similar to the high bandwidth technique, which was the more conventional technique that we have been using before. This also works with hip resurfacing arthroplasty, which you can see here, and we get similar acceleration time, so you can literally reduce your acquisition time. By 60% down to 40%. It works also with ankle replacement here, as in this case. And so I think complex sensing in the 2D world is especially suitable for for see Mac, because there's a lot of sparsity. The additional facing coding steps are very suitable to accelerate with compressed sensing because the information in there is already sparse. Now the question is, can you use that technique to achieve metal artifact reduction at three T similar to 1.5? And so here's a comparison that shows that we presented at the SM meeting some years ago that you can actually do that. So with that compass sensing acceleration technique, you can accelerate in afford more phase encoding steps at three T then at 1.5. In this example, shows that you can achieve similar image quality at three T, which is the bottom role compared to the 1.5 image quality at the top row. Now the the protocol of the four sequences takes 20 minutes at three T and 16 minutes at 1.5, so it's a little bit longer at three T, but not terribly longer. But most importantly, the metal artifact reduction is almost the same, and this is especially helpful for imaging centers that have mostly 3T scanners and do a lot of MSK and neural work. Here are some examples on how this new technique is helpful, so here's a factual image of a knee replacement with the old high bandwidth technique and you can see that there is still some residual. Metal artifact that is obscuring the bone implant interface, and if you apply that SEMC algorithm you can see that there is an osteo lysis there that was obscured before. Here's the patient that was unable to bear weight after knee replacement, you see a lateral radiograph here of the knee and the question was where the arrow is. Is there in osteo lysis we did an MRI and we saw there is probably no ostial lysis, but we had some obscuration of the posterior part here where the yellow arrow is. And if you apply the cimc technique you see that the ostial lysis is actually in the back, not in the front where we suspected on radio graphs. And this was the reason why the patient was unable to bear weight there. Another helpful application of their teammate technique where the metal reduction is is available is in children. For example menu have osteotomies or he refer more head and neck screw, and these at the top role make artifact at the femoral head and the question is, if there's Austin necrosis or collapsed Lego Strip City, and when you apply the see Mac algorithm, you can see at the bottom that this was all artifact and that the bone integrity is actually preserved there. You can also apply this to detect tumor recurrence in the presence of metallic implants. So at the top role you can see this is the upper arm. This is at the ulnar level and there's a little bit of artifact left and where the yellow error is. It actually obscures the recurrence at the bottom role you see there's tumor occurrence here on the right hand lower side is a contrast enhanced subtracted image and you can see this contrast enhancing tumor there that was obscured before. You can also do shoulder implant image in here and here you can see that the high bandwidth technique nicely reduces the metal artifact, but at least. It leaves some. Residual artifact that actually obscured the thermal. The human shaft here when you apply the C mic technique, you can see that there's a bone maritima here and then there's some osteo lysis here with the scroll erotic line, which is likely assign of micro motion and stress reaction. Here one question is how about see Mac versus CST Mic? Do you really lose? Do you really not lose anything when you apply compress sensing siemek when you apply? Can presenting acceleration on the C mic technique? And here's a comparison study so. Up top are the. Regularly accelerated C Max images and at the bottom of the conversent. In C. Max accelerated images and you can see the top part takes 22 minutes to to acquire Corona or sagittal PD and stir images and with cover sensing acceleration it takes 5 minutes each for each sequence and you can see that the comparison is actually quite similar and that is likely because see Meg is inherently slightly blurry because they have been combination algorithms and there so they combine the different phase encoding. Steps and it already causes a little bit of blurriness and come percent ING is very well tolerated in this in this combination. So eightfold acceleration is possible with compressed sensing. If you have a suitable sequence like see Mac and when you compare that to the unaccelerated technique you can collapse your acquisition time down to 12.5% of the original. Now let's move on to 3D acceleration, so this is. Basically, when we talk about 3D imaging, but we really would like to have is isotropic voxel size, that means the voxels are the same size in each direction and we can slice and dice the images later. Like you know from see T from thin slice ETS. You can do your own coronal, sagittal, or even oblique or curved planar recovery formations there. So isotropic voxel sizes. Basically, the paradigm that we would like to adhere now there's a lot of pulse sequences to choose from when it goes to 3D imaging, but we like to choose is Turbo fast spin echo techniques because. Those are usually the contrast sequences that are. Best for MSK imaging. Now if we do postcontrast imaging the vibes and lava and Tiger sequences are also very good and in some other instances desk for example is also excellent for cartilage. Now the goal is that the 3D images that we're creating and then later slice and dice in different directions should look as. Similar as possible to the 2D images, so up top is separately acquired 2D TSC images and at the bottom is reconstructed 3D images from one parent data set and you can see that with the new techniques there rather similar and. Assume similar contrast. Resolutions now what can you do with those 3D isotropic datasets? So they're very good to see small anatomic structures. For example, here is an image of the ankle, and where those small arrows are. This is the medial collateral ligament, which is which consists of three different ligaments to TBO navicular the TBO spring in the tibia calcaneal ligament, and it's hard to see these with conventional 2D images, but with 3D images we can actually see these ligaments. For example, here is the Talla calcaneal. Ligament here and it's quite surprising sometimes how these actually look, because we're not. We're not used to see those small structures. Here is one of the deeper layers of the medial collateral ligament here. And you see, you can see that you see can see individual factors here. Connecting the tailors in the calcaneus there. OK, what what else can you do? You can do what's called curved planar images so you can trace. Tendons that are oblique and unfold them later to show a curphey structures like the tendons in one plane. So for example here the enter tibial tendon, the extensor. The extensor tendons at the end code the parents. As long as parenting is. Brevis tendons poster, tibial tendon, but also at the bottom, rolls the medium in the lateral planter nerves, and so this is helpful to see structures in one plane that you usually would not be able to see in one plane. If you acquire that into 2D acquisition scheme. So we have been able to accelerate to the TSE in One Direction to fold for along time Selma to the acceleration of the two DTC technique. However, recently a fold acceleration. Became available because when you think about it in the 3D World, there are two phase encoding directions. 1 Incline, an one into partition or the slice direction. So that means that you can apply the parallel image in acceleration in even in either direction. So this was the one for the acceleration and we're skipping one phase encoding direction in the inclined. Direction, but you can also do that in the through plane direction. And if you shift that like this with a cabaret near pattern, then you can retain your image quality and you can collapse the image acquisition of this 0.5 isotropic data set to under 5 minutes, and so this makes it clinically feasible, and then you can acquire high resolution datasets in under 5 minutes in a clinical setting. Now if you apply caprinia versus the proper technique which you can also do bidirectionally. The kabrina technique is superior because it exploits coil sensitivities to better advantage and also overall has a better is a better algorithm to retain signal at the top row. You can see there's more green in this image then at the bottom, which is a crop of space or cry, but three DTC and making the point that there is more signal in the Kuiper hanya technique. Now it's about 12 to 13% more signal. Which is which is quite a lot. It probably translate to 25% acquisition time savings. Now what works clinically and how can you use that? So we devised 10 minute isotropic sports imaging protocols for the need. For example for three T and 1.5 T and you can combine them with an auto online technique and order online technique is a scout that applies the prescription automatically and you can acquire that in, under or around 10 minutes. At 1.5 it's a little bit longer at three tiers, below 10 minutes, and this is how this looks in practice, so this is like a 10 minute isotropic 3D orthopedic knee protocol and you would acquire a PD. An fat said fluid sensitive sequence. For example in the sagittal direction, and what because their isotropic. You can flip them. Into different directions. Now if you add the auto Line Scout, which you can see here, the protocol will consist of the order Line Scout which will be acquired 1st and based on artificial intelligence algorithm it will analyze the Scout images and prescribed your 3D slices automatically. So this makes it to 1 button push protocol and this is all you have to do. Place the patient in the scanner, push the button and the scanner will scan the patient on its own. Now, once you have your two isotropic datasets, you can later on your packs viewer, flip those satchels into axial in Corona is important, the image quality is being retained, but if if you use one of these new 3D sequences, this is quite possible now, and so here's one example on how this looks. One point 1.5 versus 3T. So I 3T we are forced to scan with a slightly lower spatial resolution to stay within the 10 minutes. We use PD contrast. There's alot of intrinsic signal and the difference is strikingly little. But when we go to T2 fat set the resolution at 1.5 is slightly slower. But since we're usually using this as a signal sequence and read the morphological information. Of the PD images, it's not such a big problem, so here's how this works in practice. So this is from our packs viewer here, so this is independent of the vendor of the Mr Scanner, and so these are the isotropic 3D datasets here and so. There, there coupled with scrolling so they're scrolling together and after scrolling back and forth. You can apply predefined hanging protocols and the predefined hanging protocols will flip. The same data set, for example in the two by three arrangement, like here with Axial, Satchel and Corona images, which are linked together like in this case. Now, because of the isotropic nature, you can not only you're not only restricted to axial, sagittal, coronal. You can also rotate the images. For example, you can rotate the images to the to the plane of the enter cruciate ligament which were showing here, and sometimes we have been acquiring Thatcher oblique images where we try to align sagittal plane to the enter cruciate ligament. You can see here with the 3D images you can post process that and you can even see in this case the intermediate poster bundle. Off the intact. Enter crucial ligament so this is 1 case that shows the isotropic nature and advantages of this technique. Now here's another example with the PAX viewer that we downloaded from the Internet. It's commercially available from a third party vendor and here you can see that the data set displays quite well. Here's an enter cruciate ligament tear with a poster medial meniscus tear here. And the images are also linked together, so this scroll together. In here you can see the post your meniscus meniscus tear, which is almost complete. Here's some more examples from our goney 3D study, so up top. If the 20 minute protocol an at the bottom is the two D3D protocol and you can see here that there is quite similar image quality, you can see an ACL tear in the medial collateral ligament tear here. Here's another ACL tear at 1.5 Tesla with similar image quality between the 20 minute standard protocol and the 12 minute 3D isotropic data set protocol. At 1.5 T. Here's another comparison at three T you see can see a medial meniscus tear outlined by the arrows here. And here's some more examples here. There is a prepatellar Bursitis post, your meniscus tear. The lateral meniscus tear and intra articular mass which turned out to be a fibroma. This also applies to other joints like the ankle here, so on the left hand side is the 2D and on the three on the left on the right hand side is the 3D comparison. You can see here an ATF eltar here on the PDM titeuf exit images. Here is the retracted ruptured biceps tendon that has ruptured beyond the trend line here outlined by the yellow image arrows. This technique can also applied be applied to pediatric patients, so here is a young athlete that has a discord meniscus with a tiny tear outlined by the left hand left Lower Arrow and you can see that even small tears like this you can see with the 3DS attribute data set. We did a study, evaluated the accuracy compared to AFR OSCA P and we found that the accuracies in the high 90 percents for meniscus and ligamentous lesions an in the 80 mid 80% for cartilage lesions. Here's an example on what on the smallness of meniscal tears you can detect with the isotropic 3D datasets. And here's a small free edge meniscus tear that has a nice correlation on Arthroscopy. So what about Ivy contrast? Can you still use Ivy contrast? Yes you can. So for example, you can combine that with one of these gradient echo pulse sequences that can become presenting accelerated, and then you could add a pre and post contrast sequence like here. In this case an can be subtraction images in this unfortunate patient who has a. Abscess in the distal tibia with subperiosteal. Abscess exclusion along Difices here. Here's the patient with the intra articular mass and you can see that the contrast enhanced image on the right hand side was acquired quite fast, but it is only for contrast purposes, showing that this mass is actually enhancing. Here's another patient with a synovial sarcoma of the elbow laterally next to the Radiohead and you can see a combination of her T1 weighted 3D sequence T2 fat set and then pre post and subtraction. Contrast images showing the contrast enhancement of this lesion. So far, looks adoration is readily available. It is proved in Europe and the US and available on the new generation scanners these days. How about 6 full acceleration? So 6 hold acceleration is possible if you apply the conference sensing algorithm to three DTC as we have shown before, and here's a comparison study that compares two DTC with 3D TSE that has been accelerated with Converse tensing to A6 fold acceleration, and you can see that the images here achieve clinical quality and showed slightly higher spatial resolution and then the 2D top. And you can see for example on the actual image is this flipped fragment of the meat of the medial meniscus tear outline by the black arrow? That is more difficult to see on the on the two DTC images. This works quite well with fat suppression, so these are T2 fat SAT images and you can see you get a nice fluid contrast here and also a better visibility of the flipped meniscus fragment here on the left lower hand. Which is here corroborated by the Arthroscopy image. This has also been applied to the cube technique, which would lend to me by Doctor Chowski from Wisconsin, and they have also been able to substantially accelerate a cube with the cover sensing technique here. Now, how about applying this technique in the shoulder so the shoulder is difficult to image because there is a lot of tissue that is surrounding the shoulder that can wrap into your two face encoding directions. So one way to acquire these would be with the axial plane prescription and then 100% oversampling slice encoding direction, and in this case is A is an example that this cover sensing space technique can be used to get isotropic images of the shoulder in quite high image quality, so these are .5. Isotropic datasets off the shoulder with no Rep artifact and you can see that you can slice and dice those in different directions an align them in accordance with your with your usual imaging planes. Here the PD initi two fed set or PD fed set that image image contrast. How about the risks or the risk we would like to have higher spatial resolution, for example, 0.3 isotropic data and so this is difficult to achieve. But with that hyperemia or intermediate waited cover sensing technique, this is actually doable, as can be seen here. In this example, in about 5 minutes with a 16 channel transmit receive coil here. How about further accelerating? We just said sixfold acceleration is available with cover sensing, but we would like to go even faster. So if we have to go faster it's not a limitation of the sequence. The Hibernian conferencing space or 3D TSE sequences are able to accelerate ninefold for council. For example, three and three in each direction. The problem currently is that for example, the coils are not made for that acceleration yet. So if we were to construct. A higher density coil, which is this which this prototype shows here that we are currently evaluating this is the 20 channel identity nikoil you can actually see on the left hand side is the two by two acceleration and on the right hand side is a three by three acceleration, which the geometric properties of this coil do allow. And you can see that you can collapse your imaging times by almost another 50%, so the ninefold acceleration for example is possible, but it need certain equipment, but from a sequence POV this is actually. Readily available, so if you were to ninefold accelerate your 3D images with a with a dedicated coil, then you could collapse your initial imaging time from 100% down to 11.1%, which is quite substantial. Now can we go even further? Yes, we can, so we could try to accelerate tenfold with the compass sensing technique, and that is possible. That technique is not limited to A to any acceleration. However, if we apply the tenfold acceleration. We do get a drop in image quality, however the question then is can we still use this technique for patients where we already know that cannot tolerate along merging scans? But we would still like to have the information. For example, in the Ed where where we say a 9 minute 92nd scans would actually be helpful and so this is one study that we tried here. This is comparison to DTS see on the left hand side, the 3D TSC that is acquired with five minute sequence and now here a 92nd ultrafast. Space sequence also with .5 azeotropic datasets but highly accelerated. You can see that the image quality is slightly different in appearance, however the meniscus tear here is still quite visible. Here's the axial images of the same patient and you can see there's a displays or distracted post your root here here that you can still make the diagnosis on the 92nd image. Here is the poster router again in common comparison, and although the image quality slightly different, the image, the diagnosis. Made by all the observers in this study, now he is a patient that has cartilage, loss of the medial patellar fiset here, which is also visible on both sequences. Give the patient that had an ACL reconstruction and has now a medial meniscus tear and you can see this truncation of the medial meniscus is visible on all three sequences, including the 92nd isotropic pulse sequence. So thankful acceleration is possible. It currently comes with a slight decrease in image quality, however that does not mean that it comes with decreasing diagnostic accuracy, so for certain patient population I think this can be a helpful technique as well. Now in the last part of the talk, we're going to talk about advanced post processing. So since we're now able to acquire these high resolution isotropic datasets, what can we do with them in the post processing? We know from see T that's a trouble. Datasets were very helpful in post processing 3D. Using 3D techniques for post processing now I would like to make one comment sometimes, so this is a very good sequence for cartilage and it has been used quite abundantly in the cartilage research community. When you compare the new isotropic datasets datasets from the TSE from the new 3 DTC acquisition techniques with the desk, then you can see that there is an increase in contrast. For example, for the sender image to two DTS, the image you can see. The meniscus tear better and you have a slightly better contrast to noise ratio between your suppressed bone marrow and the cartilage. There also of the fiber cartilage of the meniscus with the fluid that is in that air here. Now you can do automated Carlos segmentation for example, so this is the 3D type in this space, isotropic data set an when you apply a smart algorithm like here with deep learning that learns how to segment automatically the cartilage, then you can use that technique to cut out the cartilage and make like 3D models like shown here. You can also use a technique that called cinematic rendering applied to those high resolution azeotropic datasets and here is the patient that had a cap injury. Have a small hematoma layering along the fascia is here and you can see that when you apply the cinematic rendering technique you can get a better realization of that hematoma. Here you can even make it in a 3D technique with and overlay technique to isolate the hematoma. In cutaway the muscles partially. And here's this case, again with a tibial Abscess along the faces and the subperiosteal Abscess, and you can use the cinematic rendering technique to actually show the skin surface, which is showing quite a swollen ankle here or 4 foot and then you can use, uh, the 3D techniques in the on the right hand side to show give a 3D appearance of the subperiosteal Abscess and also of the hyperemia of the vessels surrounding the ankle. Here some examples of post processing or knee injuries on the left hand side isn't. ACL and enter cruciate ligament tear here in the mid substance in the center images and insertional patellar tendon tear and on the right hand side is a bucket handle meniscus tear. We can see the handles displaced medially into the intercondylar notch here. You can also try virtual across can be like in this case, so this is a patient that has a horizontal meniscus tear of the post to your horn and you can see here you can. Use those 3D models to actual virtually flies through the joint here, and this is afforded by the high spatial resolution of those 3D datasets and also the high contrast resolution and the high contrast resolution is sufficient for the cinematic rendering technique to actually use the MRI datasets rather than the see T datasets to make those 3D models, and so 11 application could be that surgeons could train or even practice. The Arthroscopy beforehand or it could be a nice patient communication tool to better visualize the injury in that knee. Now the last part I would finish. I would like to finish up with synthetic MRI. This is an exciting new technique that can also be used to accelerate imaging. So what is synthetic tech MRI? Synthetic MRI is a technique that uses a fingerprinting or quantitative MRI techniques and it's basically acquiring the T1 and T2 value of every pixel in an later calculates them are images rather than acquiring them separately and what it allows you is it allows you to no longer have fixed the RTS aunties like you can see. Here, but you can shift them so with synthetic MRI you can freely choose your T artizen inversion times like you can see here in these examples. So this is quite exciting and with two with the parental data set you can later reconstruct your images and calculate them with the desired 30 NTI and even more so synthetic MRI has been included in the test sequence here by my colleagues from Stanford an in addition to the death sequence that can acquire T2 Maps in parallel and get the quantitative information as well. This is also been. Applied to TSC technique with the crop bikini algorithm and in addition to your regular PD waiter Turbo spin echo image, you get an additional T2 map here for free as well. Now there's something that's called complete synthetic MRI that's no longer restricted to T2 mapping, but you also get a T1 mapping with it, and based on that you can calculate T1PDT2 Maps. R1 and R2 Maps, and based on those data you can later calculate your T1 image or T2 image PD image and you stir images which can be helpful if you can acquire the T2 and T1 mapping in a fast pace, then later you can reconstruct those images and have a time saving. So here's a comparison between conventional MRI. In synthetic MRI at the bottom you can see that you can achieve similar images, image quality and also diagnostic accuracy. Now here's a further step, that of a project that we were able to present this year at the isomer meeting which is a 5 dimensional synthetic MRI. So basically you're quiet, you're acquiring the synthetic MRI technique, but in addition to TR TNT water excitation set suppression in the medic transfer magnetic transfer contrast, so also. Variable, so MRI is no longer restricted to these parameters, but can be freely adjusted. And in this case this was done at 1.5 Tesla and the entire data sets were acquired in six minutes and 46 seconds. Now at three Tesla that can be acquired even faster. And if you apply all the acceleration techniques in addition to what we just discussed, you can go even further and that would be the hope for the future that you can go even faster with synthetic MRI. Now, in conclusion, we have reviewed a multitude of acceleration techniques that exist for 2D and 3D MRI in the 2D world there is 2 two. Eightfold acceleration, already available, and it consists of parallel imaging simultaneous multi slides encompass sensing techniques and in addition to deep learning and artificial intelligence techniques we can probably go even further than that. Very soon in the 3D World we have already tenfold acceleration shown. It's also consisting of parallel imaging, Kiper. Any acceleration, compare sensing acceleration and probably even more than 10 fold acceleration when we apply the currently being the algorithms with deep learning that are currently being developed. So it's for us to use those techniques and apply them to our current MRI protocols and see where this best fits in your practice pattern and where you can accelerate your MRI imaging protocols and hopefully retain the number of sequences and the image quality and make the diagnosis similarly accurately. There's a lot of people that help us here at Johns Hopkins to do research and develop these acceleration techniques together with our reduced search field, industrial partners, and I'd like to thank them all, and I'd like to thank you for listening and thank you very much. Doctor Fitz, thank you very much for sharing with us. Your personal and clinical insights about acceleration techniques and musculoskeletal MRI. So Doctor Fritz let me start with question here about whether or not you're using AI for any of the rapid MSK MRI techniques, and if so, how did you implement it and what are the benefits of AI use in MRI MSK. Yeah, that's a very good question, so currently clinically we're not yet using AI or deep learning algorithms, But if you go to the meetings there, being developed everywhere an I think one really good application will be for image reconstruction like the examples I showed in the lecture that will lead to me by. You so I think there is a substantial benefits for using smart image reconstruction algorithms that can. Reconstruct high image quality based on highly accelerated images, and I think this would be one application that would be really beneficial. Alright, well thank you for that another question. Here comes in from the audience. Member speaks about MRI. You know being played with various tradeoffs of 1 degree or another. In your experience, what are the most significant tradeoffs in using these accelerated techniques? So sometimes with the acceleration techniques, it's a, it's a. It's a function of how fast you accelerate, so at some point the image quality will slightly decrease or even further decrease by the, which is basically scaled by the acceleration factor of the acceleration magnitude. So I think at some point the image quality gets less an we have to figure out how much we can sacrifice for the speed to still make accurate diagnosis, but I think at the at the fourfold sixfold acceleration there's very little tradeoff between image quality and speed. And I think also by using high density coils we can retain the image quality in a diagnostic accuracy completely. But at the clinical level, I think every practice will have to apply these acceleration techniques and find out what they like best and what they can work best with. Alright, well thank you for that. Do you feel that? I guess in the future the technique of compressed sensing cimc images will partially or completely replace high bandwidth images. So yes, I know I think it depends a little bit on the structure to be image. For example, for the soft tissues around arthroplasty implants, we're substituting high bandwidth images lift, and you compress, sensing simank images. Now, that technique is still an FDA approval stage, but once it's out, we will replace those. If you're looking for nervous around implants, we will still stick with the high bandwidth technique because the inherent blurring of the cement technique, which also is shared by other. A metal artifact reduction, advanced techniques. Well, obscured the fine detail of nerves, so it depends a little bit on what you are looking for. But for implant imaging in the bone and the soft tissues around implant, my prediction is or. Our practice will be to replace the high bandwidth images with the conferencing siemek images since their time neutral in this fashion. Alright, thank you for that Doctor. So how do we go about changing our MRI of the knee protocol to implement the proposed eightfold acceleration cimc protocols, if they're using more familiar slash, traditional femc pulse sequence? Right, so there is not much change. So basically you need the pulse sequence file and then you can translate the same protocol. The same imaging parameters to the compressed sensing sequence an what it will do to acquire it will acquire it much faster, but the image parameters will be retained, so there's basically it's going to. It's going to be the same. OK. So here's another question that comes in how much time do you save using the complete synthetic EMAR approach that you mentioned, and if So what software is needed for this technique? Right so so currently I would say it's in the time neutral stage. It makes more imaging acceleration to really save time, but I think the current research is more proof of concept that this can be done and then as we go forward adding more acceleration techniques would hopefully make it faster. However, conventional imaging also gets faster in the running a race currently, but one thing to keep in mind is that there's a lot more contrast that you can get out of synthetic MRI, which could be helpful as well and you can get the quantitative. Data as well. So if you would acquire all this separately, there is already a net gain from synthetic MRI. Alright question here rather interesting one. Actually as we talk about after scans, do you think that of scans keep getting faster and faster? At some point the reimbursement for Mris will be reduced since it doesn't take as long to complete the MRI scan. Yes, that's a very interesting question. I think that will reduce anyway, so it's probably on us to get faster and faster. But the question is a good one. If we get faster and faster. If this will drive reimbursements even further down, I would hope not. So, but what I know is it will improve patient comfort and also will increase the availability of RMR slots and so that's a good thing from that perspective. But many people would predict that reimbursements will reduce anyway, so I think we're basically bound to faster scanning. Alright, just a couple more questions here and approaching the top of the hour regarding 3 dimensional. Tipard India space? What are the clinical benefits to the patient? And specifically, how can we update our current protocols to the five minute acquisition time? Right so the. The advantage would be if you're using a 20 minute protocol in your acquiring 25 minute 3D azeotropic data set protocol. Then you're saving 50% time. Another advantage could be that the spatial resolution of the 3D datasets may be much higher than the 2D datasets, and so structures of small structures with fine detailers are seen better. Like I showed in the ankle images, so those could be advantages. How do you get the sequence so you basically have to talk to your vendor? They have to install the sequence on your scanner and for that sequence. My last knowledge is that this is actually a free upgrade and you can get it on your scanner if you ask them. But for that Doctor, couple final questions here, somebody asks about software applications. What software application is performing the advanced post processing for you? So the cases I showed was done with cinematic rendering. Which is which is a vendor term. A similar technique is global illumination. I think there's several vendors now that have similar 3D techniques that have create those photo realistic images, but the one I showed was cinematic rendering software. And let's see here, somebody asks whether or not these techniques slash protocols can be used in cardiac imaging, yes, so I'm not a cardiac imager. Eminem, SK image sure, but cover sensing is avidly used in cardiac imaging, so a lot of the dynamic. Cardiac MRI imaging protocols already use compress sensing, and to my knowledge is already approved in many countries of worldwide. Alright, well it looks like we've reached the top of the hour here. Doctor Fritz and I personally would like to thank you on behalf of all of us here at applied radiology for taking the time to put together this talk today. And I can also certainly speak on behalf of the folks that Siemens Healthineers who makes today's event possible. I'm sure they thank you as well for your time and effort, so thanks again Doctor Fritz for joining us today. Thank you for having me. Our pleasure, our pleasure, and as always we want to thank you in the audience for joining us today and for your continued support of applied radiology. We look forward to seeing you online for another applied radiology expert forum webcast in the future.

Medial Meniscus Tear with Tibial BME and Medial Meniscus Tear 3D MRI: 1 Volume = 3 Standard Planes What is Synthetic MRI Performance in Pediatric Knee Conditions UDIOLOGY Accelerated CAIPIRINHA 3D SPACE TSE with 24-Ch HD Combined Use of SMS and Parallel Imaging Superficial Layer: Tibiocalcaneal Ligament 10-min Isotropic Sports MRI Knee Protocol 3D Cinematic Rendering with 3D TSE MRI 2-FoId Acceleration with Parallel Imaging Compressed Sensing Accelerated CUBE 1.5T: 5-min 2D SMS-GRAPPA Knee MRI Fully Automated 10-min Knee Protocol 3D MRI: 1 Volume = 2D Compressed Sensing Acceleration 3T: 5-min 2D SMS-GRAPPA Knee MRI Isotropic 3D CAIPIRINHA SPACE MRI DESS vs 3D CAIPIRINHA SPACE TSE Combined CS and SMS Acceleration Simultaneous MultiSIice Acquisition 3D CS SPACE: 3D CS SPACE: Asymptomatic Man 3D CS-SPACE TSE: Meniscus Tear 3D CS-SPACE TSE: Bone Implant Interface: Osteolysis 4-fold Accelerated 4-fold Accelerated 3D 4-fold Accelerated 3D SPACE TSE What 3D Pulse Sequences Work? How Does This Work in Practice? Automated Cartilage Segmention Femoral Head Cartilage Integrity 2D TSE: How Fast Can We Go? 3D TSE: How Fast Can We Go? Oblique Planar Reformats ACL Field Strength Considerations 1 Volume = 3 Standard Planes Compressed Sensing SEMAC Compressed 5-Dimensional Synthetic MRI Efficiency adds Value to MRI Shoulder Implant Loosening Scanner Generations Matter Traditional TSE Acquisition CS-SEMAC 1.5 T versus 3T SEMAC versus SEMAC versus CS-SEMAC Complete Synthetic MRI ACL and MCL Tear Tumor Recurrence ACL Tear at 1.5T 1.5T versus 3T Meniscus Tear Surface Coils Disclosures Objectives Osteolysis 3D VIBE 3D TSE Asymptomatic Man Sensing SEMAC Curved Planar Images 3 Standard Planes 3D SPACE TSE CS-SEMAC SPACE TSE 10 min Isotropic 3D Orthopedic MRI Protocol 10-fold acc. 3D Compressed Sensing SPACE 14 min MSK Neoplasms — Synovial Sarcoma How does this work in practice? TR, TE, are no longer static Synthetic GRAPPATINI MRI Virtual Arthroscopy Acknowledgments Conclusion Thank you! The Goal Partial-thickness Tear of the Patella Tendon Origin 10-min High-Resolution Isotropic 3D TSE Ankle MRI 12 min 3D Mass Protocol With Contrast — Focal PVNS Machine Learning-based Image Reconstruction CAIPIRINHA SPACE versus 2x2 GRAPPA SPACE CAIPIRINHA SPACE versus GRAPPA SPACE 10-fold Accelerated 3D Compressed Sensing 3D SPACE TSE 4-fold Accelerated SMS-GRAPPA TSE: 5-min 2D Knee MRI 14 min 3D Isotropic Orthopedic MRI Protocol With IV Gd Prepatellar Bursitis O Siemens Healthcare GmbH, 2019 PDFS PDFS cor GOKnee3D - 10 min GOKnee3D 10 min Compressed Sensing GOKnee3D-1.5T - 12 min Signal-to-Noise Ratio Maps IW STIR PAT2-SMS1 Product TSE (TA 2:40min) TSE Acquisition Time TSE TSE SEMAC TSE Disclaimer PD sag PDFS ax GrappaTlNl sequence (TA 6:30min) ax Tl cor L" PDFS cor Acceleration Techniques in üSß MRI: Math" Mtka Accelerate Acquisition through pseudorandom Undersampling Accelerate Acquisition through pseudorandom Under-sampling T2FS sag Karen Horbn Please note that the learning material is for training purposes only' Please note that the learning material is for training purposes sag SEMAC TSE Intermediate-weighted Fluid-sensitive with Fat Suppression retain or SNR High • Apply acceleration techniques to create rapid • A multitude of acceleration techniques exist for 3D CAIPI- PD/T2FS 3D Acceleration Techniques in MSK MRI: Acquisition Time improve image improve ISMRMR 2019: Research Support: Siemens AG, BTG International, TEE 73 TEE42 TE-U 2105 TEE21 '326 TEE 78 N 74 TE-26 3158 3684 70S TRE270 5263 8421 toss 42' 1 0526 78 6-316 8947 Hhanshu Bhat Jonatlan Lewh Jonatman Lewh Synthetic PO-weighted T2 Map PO-weighted T2SPAlR For the proper use of the software or hardware. please always use the Operator Manual or Instructions for Use the proper use of the software or hardware. please always use the Operator Manual or Instructions for Use Improving Efficiency, Improving Patient Experience quality 1.5 T 3D TSE TSE Click Your Fat Away 3D TSE 2D TSE VIBE 100% Unaccelerated Unaccelerate? This Program Has Been Sponsored By compensate issued by Siemens This is to be used training Weqas Wmas Mabed Wes Gibon clinical MRI protocols for the musculoskeletal 2D and 3D MRI The herein illustrated statements made by Siemens' customers and physicians are based on their own and discrete TSE Laura Fayad Laura Fayed 1.8 -2.1 x higher SNR 1.8 2.1 x higher SNR SPACE Microsoft Corporation, Zimmer CAIPI-SPACE Advanced Post- LAVA How to further for decreasing Abstract: 1218 material only and shall by no means substitute the Operator Manual. Any material used in this training will not be Improving Efficiency, Improving Patient Experience There may be content within this presentation that is research opinion. The speaker is responsible for obtaining permission to use any previously published tables. The opinion. The speaker is for Obtaining permission to previously published The Acquisition Time acquisition time acquisition uime Unaccelerate? Unaccelerated ONE Li Pm Lars Lars La-Jer ShWani Ahlakt updated on a regular basis and the Of the and hardware updated on a basis and does not the latest version 01 the software and The distribution o! this its is not permitted without written The distribution Of this training its is without express experience Question & Answer Discussion speaker is also responsible for obtaining permission to reproduce any photograph showing recognizable persons. Biomed, Depuy Synthes, QED system 2D TSE MRI can be 2 to 8-fold accelerated with Experience 5 min o\ooe Unidirectional PI authority. Offenders will be liable for damages. Isnoij091ibinU only and/or not currently available as product in the U.S.; the at the time Of the training, at the time Of the at the time 01 the Pad Khan* Hdko Mey« •0 00 00 '0 00 00 Hdls Poter PATI -SMS2 -SMS2 PAT I-SMS3 PAT2-SMS1 PAT3-SMS1 SIEMENS The statements by Siemens' customers described herein are based on results that were achieved in the customer's Dimensions: Johns Hopkins and Siemens AG Patents: R±ert Nedng R±ert Sedng future availability of these products/offerings cannot be The Operator Manual Shall be used main in particular safety information like warnings The Operator Manual be main in particular for safety information like unique setting Since there is no "typical" setting and many variables exist there can be no guarantee that other unique setting Since there is no "typical" setting and many variables exist there Can be no guarantee that Other Since there is no "typical" setting and many exist there be no that Other Isotropic voxel size! PI, SMS, and Compressed Sensing techniques • Implement acceleration techniques into clinical contrast? T2 Tl pre Tl c+ 4-fold 2-fold 50% Processing oo 40-fold 20-fold 80-fold 10-fold Unaccelerate? accelerate? Unaccelerated 25% leads to incoherent artifacts Jan Fritz, MD, PD, DABR Experience Efficiency STIR PD Hosted & Moderated by and cautions. customers will achieve the same results. 1. TR . TR Jan Fritz, MD, PD, DABR Jan Fritz, MD, PO, DABR STIR CAIPIRINHA decrease need of MEDIC retain number and McFadmd Kober 1:05 min 4:26 4:28 guaranteed. Please reach out to a local U.S. contact regarding arantee Scientific Consultant: anesthesia MERGE Siemens AG, GE, QED, BTG, 100% 1:05 MPGR types of sequences 4:28 2D TSE-3T- 10 min Kieran N. Anderson 20 TSE - 1.5T - 20 min Associate Professor Of Radiological Sciences Healthineers Hosted & Moderated by 2D TSE 3D TSE practice 2. TE 2D TSE 3D CaiPiSPACE 3D CaipiSPACE CS-3D-VlBE Associate Professor Of Radiological Sciences Iterative Reconstruction: 2D TSE MRI can be 2 to 80-fold accelerated with 3D TSE MRI can be 2 to 10-fold accelerated with Note: Some functions shown in this material are optional and might not be part of your system. The information in Dhgxhg Some not in to Some mentioned) not in Due to Some (here mentioned) not available in all Countries, to MPN Peg Cooper Peg 16 ch Rx Ankle 15 ch Knee 16 ch Rx Wrist/ Hand 16 ch Rx Hand 16 Rx Hand Jan Fritz, M.D., P.D., D.A.B.R. Jan Fritz, MD, PD, DABR Director Of jntewentionaf MR Imaging CAIPIRINHA 30 SPACE TSE CAIPIRINHA 3D SPACE TSE Director Of Interv'entional MR Imaging Director Of Interventional MR Imaging Director Of Intewentional MR Imaging Director Of Intewventional MR Imaging Director Of Inter.'entional MR Imaging CAIPIRINHA 3D SPACE TSE CAIPIRINHA 30 SPACE TSE 3D SPACE TSE the availability and Regulatory status of the features/offerings Minimize a cost function that Director Of Interventional MR Imaging 5 min this material contains general technical descriptions Of specifications and options well Standard and optional 2 min regulatory reasons their availability cannot be guaranteed. Please contact your local Siemens organization for further Kieran N. Anderson Alexion Pharmaceuticals 3 T ima es 4 x faster 3 T ima es 4x faster Group Publisher 3D CS SPACE: 30 SPACE IW 30 SPACE IWS 30 SPACE Cesar Neto John Camino John Carrino Omnis Agosdm Omnis Associate Director MSK Imaging FeYlowship Professor of Radiology and Radiological*cignces Radiology and Radiological Science Associate Diæctor MSK Imaging Fellowship Associate Director MSK Imaging FeYowship Associate Diæector MSK Imaging FeYlowship Associate MSK Imaging Fellowship GRAPPA + 25% GRAPPA + SMS Associate Director MSK Imaging Fellowship Asscciate Director MSK Imaging Fellowship Associate Diæector MSK Imaging Fellowship features that do not have to be present in individual Associate Director MSK Imaging Fellowship GRAPPA + en in individual details. enforces sparsity! with PI, CAIPIRINHA, and Compressed Sensing • Illustrate a variety of post-processing techniques Applied Radiology Applied Ra diology RULE Cube-CS Cube 4. WE mentioned herein. Johns Hopkins University School of Total Total g:3g Total TN: Director of Interventional MR Imaging Director Of Interventional MR Imaging Johns Hopkins University School Of M.4edicine Johns Hopkins University School Of Medicine Johns Hopkins University School Of hdedicine Johns Hopkins University School Of h.4edicjne Johns Hopkins University School Of h.4edicine Johns Hopkins University School of Medicine Johns Hopkins University School of Medjcjne PATI-SMS3 PAT2-SMS1 Group Publisher PAT3-SMS1 PAT2-SMS1 Lew Schon Johns Hopkins University School Of Medicine Kaga Bæhm Day Ray Honorarium: f = llU(Fx —y)ll$+ f = llU(Fx —y)ll$+ xlll Siemens AG, GE Whole-volume, High-Resolution, In•Vivo Ratio and G•factor Superiority, and Whole-volume, High-Resolution, In-Vivo Ratio and G•factor Superiority, and CAIPIRINHA+ HD coil Il .10/0 HD Il .10/0 11.1% Russell H Morgan Department of Russell H. Morgan Depa'tment Of Russell H Morgan Department Of Russell H. Morgan Department of Russell H. Morgan Department Of Russell H. Morgan Of Russell H. Morgan Departrnent Of Russell H. Morgan Departmnent Of Russell H. Morgan Depanment Of Russell H. Morgan Depantment Of Russell H Morgan Departrnent Of Associate Director MSK Imaging Fellowship 5. MT 0.5mm x 0.5mm x 0.5 mm 0.5mm x 0.5mm x 0.5 mm decrease Increase + HD coil Russell H. Morgan Department Of Radiology and Radiological Science Russell H. Morgan Department Of Radidlogy and Radiological Science Russell H. Morgan Department Of Radidogy and Radiological Science Russell H Morgan Department Of Radiology and Radiological Science Applied Radiology 3D TSE T2FS 3D Compressed Sensing SPACE IWFS 3D Compressed Sensing SPACE • I-JhChPiques techniques applied to rapidly acquired MRI data Structural Similarity Index Differences, Of Compressed Sensing SPACE and CAIPIRINHA 1.5 T Papp Hbert Cteryt Kamedne Lasby Lasby Radiology and Radiological Science Radiobogy and Radiological Science Radiology and Science Radiobgy and Radiological Science availability of "Functionality") may not (yet) be commercially available in your country. Due to regulatory requirements. the future motion Phase Encoding 2D TSE 40 ETL TR 900 ms, TE 29 ms, ET 44, Jan Fritz, M.D., P.D., DAB.R. TR 900 ms, 86 ms, ET 33, Tlc-/c+/sub Intermediate Weighted 3D TSW MR Image SPACE over GRAPPA SPACE MRI Lateral Radiograph 2D TSE 20 TSE 3D TSE 20-fold 2-fold 8-fold 80-fold 40-fold 4-fold 3D CaipiSPACE TSE 3D CS-piSPACE TSE 3D CaipiSPACE TSE MPR High-BW PD 3D CS SPACE: 12.5% availability of said Functionalities in any specific country is not guaranteed. Please contact your local Siemens availability Of said Functionalities in specific country is not Please Contact Siemens availability 01 said THEM ALL CS-SEMAC PD Use it! VIBE VIBE Compressed Sensing T2 Benamin TIC+ Kunar 5:30 min Scan Time Nov* 5:30 min Scan Time Johns Hopkins University School of Medicine Johns Hopkins University School Of Medicine Penalty term that 6-fold Acceleration 6-fold Acceleration ISMRM Data term Spars fring Data term Spars Data term Spars 'fying Data term Spars ifying AX T2 FS iPat2 90-fold TR ms, -re 32 ms, ET 36, TR = ms, 32 ms, ET 36, SMS2-CS2 DESS Echo State DESS Echo Steady State DESS State Make MRI more Healthineers sales representative for the most current information. in Spcmscxed by: Sponsored by: Sensing 3:58 3:58 sec 4:46 sec 90 sec . Use it! • 106 106 TE 40 ms TE 39 ms GRAppA 4 ML GRAPPA 4 ML GRAPPA4 3D caipi- Email: [email protected] T2FS 3D SPACE T2FS 30 SPACE 3D caipi- 4 min TE slide Tl slide TR slide TE slide 4.5 min slide 2D TSE 3D CaipiSPACE TSE MPR 3D cs-piSPACE TSE MPR enforces sparsity GRAPPA x 2 = 1/2 acquisition time Russell H. Morgan Depa t of Radiology and Russell H. Morgan Depa Russell H. Morgan Departrnent Of Unaccelerated Unaccelerate? TSE of Radiology and 3D CalpiSPACE TSE MPR 3D CS-SPACE TSE MPR 3D CS-piSPACE TSE MPR VIBE 0.5 x 0.5 x 0.5 mm transform e.g. rransform e.g. post MEDIC = 0.5 x 0.5 x 0.5 mm PD sag PD sag T2FS sag PDFS cor TR x x NEX Tl cor 3D VIBE Shdl Kunar pre TA = 3:27 3:27 Z4, MRI 0.3 x 0.3 x 0.3 mm PO-SPACE P D-SPACE SIEMENS SIEMENS .. 1:43 Sagittal 3 mm images From medial malleolus to talus Isotropic O. 5 mm3 dataset Isotropic 0.5 datasets Isotropic 0.5 mm3 datasets STIR From medial malledus to calcaneus STIR 3D CAIPIRINHA SPACE 3D CAIPIRINHA SPACE 3D CS-VIBE FISP SPACE S: 03 min min LAVA Liver with Volume LAVA Liver With Volume LAVA Liver With T2SPAlR High-BW STIR 3:58 4:28 s FACE 4:46 90 SOC 4:46 CS-SEMAC STIR 90 Scan Time FOV 16 cm FOV 16 cm FOV 8 cm, 180 slices Scan scan S. S. W, PO ax Tl cor PDFS cor PD ax S. W, Healthineers Tl cor PDFS cor T2FS sag PD sag Isotropic 0.6 mm 3D TSE Isotropic 0.5 PO 3D TSE Isotropic PO 3D TSE Isotropic 0.5 PO 3D TSE Isotropic 0.6 mm 3D TSE Isotropic 0.6 mm T2FS 3D TSE PDax Tl cor Isotroø•c 0.6 mm DESS 0.6 mm DESS PDFS cor Successive image acquisition = high image quality, but (nowadays) time consuming FLASH Fast Shot FLASH Fast FIESTA Employing FIESTA Employing STate FIESTA Fast Employing STate Sutrer Estrer Simultaneous image acquisition — two or three images at the same time Flbpo Del ET x PAT . p . . p . . . . p . . p Tr-*Ü,. . p . . . p . 37 37 1 ST Acquisition time 5:12 min Acquisition time 5:45 min F. R E. C. F. R C. F. R. E. C. R. E. E. h E. h 70 R W. RL 30 C. 30 A u. S, YML N. B, S. J , MRI u. B, S. YML MRI B, S. YML J B, S. M. YML J MRI N. B, S. M. YML J MRI Del F , D , SS , J. 30 MRI Ray 8 2018, DC Del F , D , SS , J. 30 MRI Ray 8y 2018, DC Del F , D , SS J. 30 MRI Ray 2018, DC J. B. GK. H. WD. B. GK. H. WD. Acquisition time 5:10 min MERGE - Gradient Echo MERGE Gradient Echo MERGE Recombined Gradient Echo R, A. K. R. F. R, A. K. R. R. H, A. K. R. F. J R, H, A. K. R. F. A. K. R. FM 276 & 276 & 276 27m Stal&t AF. M, undomanvled MRA in a soning. Rwon Mod 2014. Stal&r AF. M, Quid undemanvled MRA in a soning. Rwon Mod 2014. Stal&t AF. M, Quid ot undomanvled MRA in a soning. Rwon Mod 2014. Stal&r AF. M, Quid undemanvled MRA in a souing. Rwon Mod 2014. Stal&t AF. M, under-sanvled MRA in a soning. Rwon Mod 2014. Flexible Phased 4 Channel RX 16 Ch RX Shoulder

  • msk
  • muscoskeletal
  • muscular
  • skeleton
  • msk mr
  • msk mri
  • MSK
  • MASK applications
  • iPAT
  • ipat
  • pat
  • i PAt
  • i pat
  • PAT
  • speed
  • shorter time
  • acquisition time decrease
  • faster
  • accelerate
  • excelerate
  • turbo suite
  • turbo siut
  • turbo package
  • turbo suite excelerate
  • turbo suite accelerate
  • turbo accelerate
  • accelerate package
  • excelerate package
  • performance
  • shorter scan
  • faster scan
  • short time
  • CAIPI
  • SMS
  • caipi
  • sms
  • msk image
  • msk imaging
  • caipirinha
  • CAIPIRINHA
  • mski mri
  • mski
  • acceleration
  • accleration techniques
  • acclereation
  • techniques mri
  • mr techniques
  • improving efficiency
  • improving patient experience
  • efficiency
  • patient experience