Workstation face-off liveblog
Here it is, the entry you've been waiting for -- the workstation face-off liveblog! For those new to blog readership, here's how liveblogging works: I'll update this entry continuously throughout the face-off, with my newest updates appearing at the top. Keep refreshing to stay on top of all the action! TeraRecon's up next. The presenter opens it up in a CTA protocol; the system removes the table and bone automatically as Dr. Herzog cleans up the image a little bit. Now he's isolated the vessels and is cutting away an artery, then moving to the MIPS view and windowing through to show the kidney stone. Saving and closing out, he logs on again as the "doctor," quickly reloads the image and indicates how to restore the missing vessel. Philips. The dataset loads in a few seconds. Cleaning up the data, the presenter goes deeper into the program to show the automatic segmentation to display the different vessels and the kidney stone. He adds in a few seed points -- "I don't really do this ever, I'm just trying to be fancy," he stresses -- then names the image "stone" and saves it. Going back a step, he cuts out some vessel and a chunk of kidney, his intentional error, then shows how the saved dataset can be moved around his network. Booting back up as the doc, he reloads his work -- it takes a few seconds longer than the others -- then demonstrates how to restore the vessel and kidney piece. It takes a while to pop up . . . and time's up before it displays. Bummer! GE's back. Loading the protocol, Dr. Wolff shows how the machine does all the work in terms of extracting the arteries and getting rid of the bones. The window automatically populates within five seconds or so. With no user input, the system has isolated the main arterial tree. He marks what looks like a kidney stone, then mags up the image to pivot around the stone, creating a nice view of the kidney. He saves it, then windows through the vascular view and has the computer automatically select the arterial tree he's interested in, displaying the four renal arteries on the right in a bright, colorful view. He removes a small vessel, then saves again (calling the file "junk," BTW). He exits, then boots up again to play the radiologist role, with 30 seconds to spare . . . he restores the missing vessel with just seconds to spare! Next up: Dr. Tanenbaum for Vital Images. He preselects portions of the vessel right at the beginning, then zooms in on his 3D model to display the arterial tree as he tries to remove the noise. Whoops -- he accidentally messed up the model in the process . . . Moving on, he zooms in on the renal arteries, then does what he imagines a 3D tech would do and marks the renal arteries in a 2D view. He creates his error and removes a piece of vessel, then snaps the view, sends it to the server and moves on to the physician phase. In the "reading room," he restores the workflow from the server -- 30 seconds to go! -- and clicks on the vessel to restore it in 3D. Ta da! Dr. Mendel's up first, for Ziosoft. In his aortic aneurysm protocol he lets the system automatically remove the bones and table, then cleans up the image a little bit by selecting the arterial tree. A bit of a blip here -- the system doesn't seem to want to cooperate and the clock is ticking . . . ah, there it is. Okay. Now we can see that there is no AAA, but now he needs to grow back the vessels in the patient's leg. Okay, now we're saving the image and going back to 3D analysis, window and level . . . he's picking a thick slab showing the renal arteries, then "accidentally" removing some vessels. He saves the file, then re-loads it, now in the role of the physician. He grows the vessels back, visualizing all the renal vessels. And with seconds to spare! Case 3: Genitourinary/Vascular. In part A of this case, presenters will play the role of a 3D lab technologist; the case comes through with a clinical history of "rule out AAA." The task at hand: demonstrate the normal aorta, the techniques available to you to mitigate against the noise of the artifacts, and create a single summary image for the referring physician. Edit the volume to demonstrate the course of the renal arteries, then create a volume rendering, save it and close out. Then Part B starts: the case is transferred to the reading room and you're the doc. Open the case and review it, then create a new volume rendering. Only four minutes for this one -- EEEP! Here we go . . . Dr. Mendel for Ziosoft, launching the vessel analysis tool and bring it up in the coronal view. He places his first point at the beginning of the stent, then climbs the vessel, zooming up and putting in his second and third points. Another three-pointer -- impressive! Now he's showing how to adjust the centerline, and he's done. Zip zip zip! Visage's dataset loads in a snap. Zooming in on the intracranial M2 segment, then moving through the skull and activating the tracing to extract the centerline. Here's the lateral view, indicating that the centerline is perfectly aligned with the vessel -- a quick zoom-in confirms. Carestream's up now. Fast load! We're checking out a stent on the right and occlusion on the left. Now she's giving three clicks -- one on the aorta, one on the carotid and one more up top. THREE SEEDS -- pretty smooth. The centerline appears accurate so far. Even though there's no deviation, she's showing how easy an error would be to correct. And now we get a conventional angiography view to conclude! Siemens. "I've got the feeling it's going to be four clicks," the presenter says as the dataset loads. Here's a view of the artery occlusion and stenosis. Checking out the carotid artery as we move up through the skull, we switch to a perpendicular view, which clearly shows that something's going on. The centerline's a little bit off, but fixed with a quick correction. Nice! TeraRecon . . . We're loading the case. What a stunning, colorful 3D view of the anatomy, now giving way to the MIP view of the vessels. There's the lumen. And he's already done -- now he's just demonstrating how the case can easily be e-mailed to a consulting physician. That was quick! Philips. Also takes a few seconds to load, but, as the presenter explains, there's an automatic preprocessing that normally happens, though not this time . . . this time we'll watch him do the vessel extraction manually in the axial plane. "I'm a seeder," he says, explaining that using only three or four often leads to a need to correct their positions, which we've certainly seen. And now, with 30 seconds to go, he's lost count of how many seeds he's planted, drawing laughter from the crowd. Window set and time's up! (He notes that the system had already automatically done this.) Next up is GE and Dr. David Dowe. The image is taking a bit to load w/ autobone extraction . . . okay, here it is. He's putting down four reference points for the centerline extraction, then showing the lumen view. Tracking looks good in the curved MPR. We're seeing the continuous vessel now -- "open and shut, slam dunk case," Dowe concludes! Case 2 is all about centerline extraction with a question about atherosclerotic stenoses. The task: to extract a single centerline with a minimum number of seed points. Here we go! Vital's subbing in Dr. Lawrence Tanenbaum, and he's already crackin' wise. Using his selection tool, he's rolling down and picking his first seed. Here comes seed number two, in the transverse carotid, and correcting his centerline. Extending further, he samples again, then again for a total of four seeds. Ruh-roh -- the system seems to have made a mistake, meaning he has to undo his fourth seed and work with the three he already has. (Yikes!) Zooming up, he displays his final centerline for the crowd . . . Last demo for case 1: Vital Images, presented by Jeffrey Dardinger (you may remember him from my lunch date earlier). He's loading the study -- it's taking a couple of seconds, maybe a bit longer than the others -- okay, now it's officially taking a looooooong time. People are starting to whisper -- how scandalous! Has his computer locked up? Okay, at 1:30 the study has finally loaded, and now Dardinger's gotta talk fast. 3D rendering. Skin rendering -- also taking a hair too long. He rotates the foot. Not to pick on Vital, but seeing a little bit of a delay on all processes here. 30 seconds to go, and he's showing the vessel analysis, displaying the occlusion and rotating the image to make his diagnosis right under the wire. At a request from the moderator to make the image bigger, he says he'll need another 20 seconds . . . Here comes Jeff Mendel for Ziosoft, a first-time participant. Here he goes into his hanging protocol, which has two views, one MIP, one 3D. Zooming into the foot area, we see the skin rendering and the tissue rendering. Now we can easily see the osteomyelitis as well as some incidental bone findings, including screw holes from a previous surgery. Without even removing the bone, he's showing the vessel view and rotating the foot. The simultaneous views allow us to clearly see the vessels. And the color palette is so pretty! Now Visage has the floor. With four views onscreen at once, we see the skin, then move to a close-up view of the ulcer in the bone template. Windowing in and out to see the skin and vasculature . . . on the right side of the foot we visualize the veins before going to the MPR view and angling in to closer examine the joint. In this view we can examine the occlusion, and zooming in, we can see the venous filling and then go to axial to verify the stenosis. Done with time to spare! And Dr. Meyer gets the "zoom award" for really being careful to make the images visible to the whole audience. Next up: Michalle Soudack for Carestream. She's showing the skin and bandages as well as how to remove them, then swiveling to show a spot in the ulceration and swelling around the right angle. Fun fact: Carestream's visualization is done within its PACS, not a separate workstation. Moving on to the coronal view, she shows some of the irregularity in the foot and atrophy in the muscles. In the MIP view we can see the occlusion and venous filling. And now she's establishing the best area for myocutaneous flap . . . one final close-up view with just a few seconds to go. And that's time! Anno Graser's presenting for Siemens. He says he's mostly an abdominal guy . . . a theme emerges. Perhaps these savvy vendors want us to know it's so easy a kid could do it? Now he's visualizing the skin and the sinus tract, peeling away virtual layers of foot to reveal the osteomyelitis within. Changing the preset to a runoff and removing the clip point, he indicates an occlusion, then goes to automatic bone removal. Switching to a thick MIP, he shows the occlusion and the arteries within the foot to make his decision! Here comes Chris Herzog with TeraRecon. The CT runoff template automatically removes the bone and table. Now he's magnifying the 3D image and showing the osteomyelitis and some possible osteoporosis in the ankle region. Here's the MIPS view, showing some stenosis . . . and now he seems a little lost, switching between images. Oh. He's found his way and has made his diagnosis with seconds to spare! Next up: Philips and Claudio Smuclovisky. He's doing a dual-monitor view. Oops -- looks like the patient "is not Jewish," in his words . . . panning down now to the feet, we clip off some of the bone to check out the vessels. In the 2D view he demonstrates the osteomyelitis, and checks out the vascularity in the axial view. Now he's checking out the MIP and the 3D simultaneously, showing the arterial occlusion . . . snazzy. -- Here comes Dr. Steven Wolff, batting for GE Healthcare. "I'm principally a cardiac and vascular radiologist," he said, "so the last case of osteomyelitis I saw was probably 15 years ago." He's demonstrating removing the hardware from the image of the ankle, then windowing in to show the skin and the surface ulceration. Now we're checking out the thin MIPS with the sinus track extending into the bone. To look at the vessels, he's getting rid of the bone with a click. A few seconds' delay, then a view of the vessels that we can trace down the leg. The arteries look good until the foot -- is this guy a diabetic? And then a MIPS view of the foot showing which artery can be sacrificed for myocutaneous flap without causing ischemia in the foot. Case 1 will be "skeletal vascular." The cases are very brisk -- just a few minutes apiece per presenter. This case is surgical planning for myocutaneous flap transfer in a patient with chronic osteomyelitis. The first task is to use volume rendering to illustrate the skin surface and the ulceration with sinus track extending into the bone. And we're off!