Technical Aspects of Breast Tomosynthesis Video

Thank you very much. We've been using digital breast time of synthesis for a number of years now, and. About three years ago, we expanded our practice and in the process ended up having two digital tomosynthesis units in the practice and. These units were from different manufacturers that was quite interesting to to see the comparison and both of these machines obviously had regulatory approval and both of them. Have been shown to increase cancer detection so soon after we had this example of this well, this 55 year old woman presented she had been previously well, she had a palpable lump in my left breast. You saw her referring Doctor Who sent her in for her first mammographic examination and you can see the mammograms very dense and heterogeneous. There's some mild asymmetry. There's a small stromal opacity, but this stronger pasetti really did not fit with the clinical findings. She had a large palpable mass. The clinically was highly suspicious, and this small stremmel density didn't appear to correlate. She did have a prominent, no, that was evaluated with with ultrasound, but for today's discussion more interested in the breast parenchyma. So we did time of synthesis. As I said, we've got 2 units in the practice. We did one. Used one of the systems and to our surprise, really the. Vision the expected palpable mass was not evident. The radiologist reporting this thought that the study was equivocal and even on a high resolution monitor there was no obvious mess to correlate with the clinical findings. So one of the options is that the lesion is actually not included. Maybe it was too far back, but if it is included then what else could we be missing in a lady that has got a highly suspicious abnormality? So we decided to just repeat the mammogram, or repeat the tomosynthesis study, ensuring that we did have the lesion included or the area of concern included. The positioning is actually very similar to the first mammogram, and it just happened that we repeated on our second system an as one scrolls through from lateral to medial. One can see that there is a more obvious speculated mass, an I'm just highlights that because I know these images don't project. Two well on, certainly on a high resolution monitor. One could see this speculated Mass an just highlighted image 43 of 70. Even in retrospect it is difficult to see. Mass on our the system and we've talked a little bit today about wide angle and narrow angle and different machines and so on, so I just really wanted to explore that issue today. And this is just comparing the two to examinations where you side by side and the ultrasound examination which. Bobbsey proven, which went on to be a biopsy, proven invasive ductal carcinoma Grade 2. So there are differences in the systems available and. Really, what is the implication of that? Is one of the questions I want to consider today. So first of all, let's just step back a little bit. And if we're designing a tomosynthesis system, then there are a number of constraints that limit what we can actually produce. Or designers and manufacturers are able to produce. One is obviously the total radiation dose we've talked about that today. We need that as low as possible, and certainly only to be in the realms of a standard 2D mammogram. We really can't have a high radiation exposure, the detectors have to have high quality. High quantum efficiency. We're doing multiple low dose exposures and so any individual exposure is only a fraction of the dose of a standard 2D mammogram, so there's high requirements on those detectors. High quantum efficiency, rapid readout from different exposures, and no lag, where the signal from one is bled, or Leora turning up in subsequent exposures. Time of synthesis has mechanical constraints compared to a 2D mammogram. The tube is X Ray tube is moving. They can move smoothly, but we run the risk of of focal spot blurring. So very, very narrow pulse width for our beam. So another constraint, the option is to have the so called step and shoot where the X Ray tube moves to a spot stops. We take the X Ray exposure, it then moves to another pre determined spot exposures again so that stop the step in should will take longer because of the interrupted motion and also run the risk of mechanical motion. So again lots of constraints on our design. We've talked a little bit about scan time today. And again scan time we want to keep that as short as possible because of the risk of patient motion and I'm sure your experience is the same as mine. We sometimes get motion artifact on a 2D mammogram where the exposure is a very short fraction of a second. So time for synthesis with a longer exposure. Obviously there is a risk of movement, but I think that in certainly in my experience the the likelihood of motion artifact relates very much to the experience and training of the. Radiographer Mammographer we heard about the importance that in the last talk and like I agree, 100% or cannot agree more really that the training of technical staff is important. They need to be trained well. They need to be interested. They need to be motivated. Need to have experience that needs to be able to relate to. The patient or the woman having the mammogram or tomosynthesis to ensure she understands the importance of not moving. And if you've got good staff, then motion artifact really is less of an issue. And we obviously need the ability to image the entire breast in one scan, so that can have limitations on what we can achieve with tomosynthesis. So there are multiple factors to consider all the way through from the X Ray tube. We've talked about the narrow angle an wide angle scan time, number of projections I've mentioned about the continuous or step and shoot approach. There is different motion geometry is available. One can have most systems. In fact the partial isocentric where the detector is stationary and we have the X Ray tube move. In an arc above the stationary detector, but some systems, and there's no reason why it's not possible to do a fully isocentric system with X Ray tube and detector rotate again. Differences in geometry, different implications, grids, grids will obviously decrease, scatter and increase the soft tissue contrast, which is good, but it increases exposure dose, which is bad. So again there's all compromises and and that we need to make detector type just As for a standard. Digital mammogram it can be amorphous. Selenium or cesium iodide pixel size is obviously important. We want small pixels to increase the resolution current systems anywhere between 50 and 100 microns. If we have more pixels then it's going to take longer to read out, so that's going to be a limitation. And we have less signal from any individual pixels, so it gets back to having high quality detectors. One way of speeding up the readout is the so called pixel binning, which you may have heard out. Binning really just refers to the combining the pixels. So if we combine the pixels we can read out faster, will have a larger signal from each readout, But then that will if we combine the pixels then we're getting a decrease in the resolution, so the effective pixel size actually increases. Less resolution, so again a compromise. And finally, reconstruction algorithms as incitti. It started with filtered back projection, but there are more modern algebraic and statistical algorithms we're going to hear more about that. I think later on today about the advances in algorithms and all these will obviously affect the design and the final image quality. So two of the factors that do really define Thomas Synthesis technology are really under. The two parameters are the number of projections and the scan angle an number of projections? There's really just like a standard CT scan. The more more projections we have, the less artifact, and I've just highlighted this nice example from the literature which shows for the same scan angle. In this case 45 degrees 9 projections versus 25. And one can see that there's much more streak artifact with nine projections. Increasing the number of projections decreases the artifact, and we can see the abnormality to better advantage. Time we've had quite a bit today about the scan angle and increase. Obviously increasing the scan angle increases the tissue separation. That is positive because we get less superimposition and less anatomical noise or less superimposed superimposed to tissue. And it also increases the Z axis resolution. Now all manufacturers produce images with slices 1 millimeter apart, but it's actually not true to say that all these slices are 1 millimeter in thickness. We only have a limited scan angle, so in fact, the thickness of the slice is not well defined. It's not not an isometric voxel, it's a poorly defined slice in the Z direction. But what we do know is by increasing the angle. Then we'll get effectively thinner slice so it increase or increases the xaxis resolution so it's effectively a thinner slice. So obviously ideally we'd like more projections. The scan angle will give us that thinner zed axis resolution, but then we've got to keep in mind our constraints that I mentioned at the beginning, but we really need to. The scan angle and the projections are limited by the detector performance. The need to limit the exposure dose and obviously the scan time. So it's not surprising that manufacturers have ended up with different combinations of all these factors, and it's not surprising that at the moment in the early days of tomosynthesis technology that there are differences in the appearance of images from different manufacturers. So this is just another example that came through our practice not that long ago. 51 year old woman, past history of a fibroadenoma. Again, there is a fibroadenoma which is easily seen on five megapixel monitors in the practice so low density, mass and looks benign. We did tomosynthesis and deep to the nipple there was a circumscribed nodule. Which not very evident on the 2D mammogram. So this study was performed November 2016, just happened to be on one of our tomosynthesis units, and we saw this nodule is circumscribed. It's probably benign, but in a 51 year old woman really wanted to know if it was a new finding or had it been there previously. She had had a study in our practice 18 months earlier, so this is May 2015 and it just happened to be on the alternative system and again we saw in earlier talked the differences between images from different systems and this is just another example which shows that when we look at the tomosynthesis image, we scroll through. It's very hard to detect a small nodule in that position, so it was some concern. This was a new nodule and if it's a new nodule then maybe we need to. Further evaluate it. But in fact, 18 months earlier she had a mammogram again, November 2013. This happened to be on System 2, the same one as 2016, and again we can see a circumscribed nodule. So just in summary, you can see November 2016 discrete nodule May 2015. Difficult to see November 2013 discrete nodule. So we reported this as a stable nodule and differences in the appearance due to differences in technology and technique. However, our surgeon who's not as experienced in term of synthesis, was concerned that there had been a change and she basically wanted us to do a biopsy. We did biopsy this and it was benign, but it's really just an example to show that the different. Machines and different systems are producing different images. And so I really just want to summarize by saying that there are a number of systems available. They all have regulatory approval. They all have been shown in clinical trials to increase cancer detection, but the manufacturers have used different combination of technologies and techniques to produce that image. And really the combination of what is of the factors scan angle projections and so on. The optimal combination is not really been determined. And it's still work in progress, and those differences can potentially result in different clinical outcomes. As you can see on those examples, and potentially can make comparison between click clinical trials of the clinical trials have been performed on different systems, then comparison and what is found on one system may not be directly applicable to another system. Thank you very much.

SIEMENS Healthineers • Healthineers Healthineers ECR 2017 Palpable lump left breast Breast Tomosynthesis •L X-ray Tube: Anode W or MO or Rh TOMOSYNTHESIS CONSTRAINTS TOMOSYNTHESIS TECHNOLOGY SUMMARY 51F Past history right fibroadenoma OF Number of Projections: 9— 25 Number or Projections: 9— 25 Number or Projections: Filter: Rh or Ag or Al 11/2013 11/2016 05/2015 11/2013 11/2016 Scan Time: Scan Angle Scan Time First mammogram Additional workup Initial workup Initial workup Scan Angle: 11- 11-500 OF 11/2016 05/2015 System I System 2 11/2013 Scan Time: Scan Angle: scan time. Scan Angle 4 — 25 sec •Systor I *Systo I Systo I Systom I Tomosynthesis System 1 System 2 System I System 2 Number of Projections: 9— 25 System I System 2 (7 Total Radiation Dose NUMBER OF PROJECTIONS • NUMBER OF PROJECTIONS SIEMENS Dr. Wayne Lemish XRT Motion: Continuous or Step-and-shoot qontinuous or Step-and-shoot Increased scan angle Motion Geometry: Motion (±ometry: MIA Radiology East Melbourne, Australia Detector Efficiency • SCAN ANGLE Partial isocentric (Detector stationary) Paprtial isocentric (Detector stationary) Increased tissue separation Healthineers Full isocentric (XRT and detector rotate) Full isoc&ntric (XRT and detector rotate) XRT mechanical motion Technical Aspects of Breast Detector Grids: Detector Grids:• Yes or No The scan angle and number of projections is constrained by Tomosynthesis Pomosynthesis less superimposition Scan Time: Scan Angle Scan Time Scan Angle: Detector Type: Detector Grids: Detectoq Type: Dletector Type: Detectq! Type: ase or CSI Pixel size: Pixel 50-100 11-500 less anatomical noise detector performance and Ability to image entire breast Pixel readout: Single or binned increased z axis resolution Increased tissue separation need to limit exposure dose and scan time. Scan Time: Scan Time Reconstruction Algorithm: Dr. Wayne Lemish FBP, Algebraic, Statistical East Melbourne, Australia East Melbourne Radiology 44/68 44/68 18/45 23/54 23/54 18/45 21/47 21/47