Best Oral and Poster Presentations Women in Science with Lisa Koch A publication by DAILY October 8-12
Qiang’s picks of the day (Wednesday): QiangMa is a PhD student at the BioMedIA group at Imperial College London, United Kingdom, under the supervision of Daniel Rueckert and Bernhard Kainz. (Oral 9) SwinMM: Masked Multi-view with SwinTransformers for 3D Medical … (Oral 9) Structure-Preserving Instance Segmentation via Skeleton-Aware Distance … (W-06-062) MoCoSR: Respiratory Motion Correction and Super-Resolution for 3D… (W-06-064) ModusGraph: Automated 3D and 4D Mesh Model Reconstruction … (W-06-088) S3M: Scalable Statistical Shape Modelingthrough Unsupervised … (W-06-103) Weakly Supervised Cerebellar Cortical Surface Parcellation with … “My research interest is geometric deep learning for medical image analysis. My PhD project focuses on the deep learning-based cortical surface reconstruction for the developing brain. Currently I am working on a learning-based pipeline for fetal and neonatal brain MRI processing.” Oral: Posters: For today, Wednesday 11 MICCAI 2 DAILY MICCAI Wednesday Qiang’sPicks “This is the first time I attend MICCAI in person. I feel very excited to meet excellent researchers and discuss state-of-the-art work!” Learn more about Qiang’s work! Don’t miss: Qiang is presenting today his own work: W-05-023 Conditional Temporal Attention Networks for Neonatal Cortical Surface Reconstruction Don't forget to visit him during poster session 5!
In medical imaging, Magnetic Resonance Fingerprinting (MRF) has emerged as a promising approach to perform fast quantitative Magnetic Resonance Imaging (QMRI). In the case of brain imaging, QMRI provides invaluable insights into brain tissue, aiding in medical diagnosis and personalized treatment. However, existing QMRI methods suffer from slow processing speeds. MRF provides an alternative QMRI framework to derive quantitative values simultaneously but also faces hurdles. “Previous MRF methods weren’t good at generalization,” Juyeon tells us. “Different hospitals use different settings for their medical equipment. It doesn’t work well when you develop a model using one hospital’s dataset and apply it to another. Our goal is to build a really generalizable model.” She introduces a novel concept to address this challenge: the physicsinformed decoder. Her framework, known as BlochNet, has a supervised encoder-decoder architecture, where the encoder predicts essential quantitative values (T1 and T2 values) from the input signal, and the decoder uses MRI physics, specifically Bloch equations, to reconstruct it. 3 DAILY MICCAI Wednesday Physics-based Decoding Improves Magnetic Resonance Fingerprinting Juyeon Heo is a PhD student at the University of Cambridge. In this work, she proposes a solution for the Magnetic Resonance Fingerprinting (MRF) problem. She speaks to us ahead of her poster this morning. Juyeon Heo
“We want to build a physicsinformed neural network, so we need to decide how to include this prior physics knowledge,” she explains. “We can’t understand how this deep neural network works, so we can’t manually make this deep neural network work the same as the physics equation. That’s the main challenge.” The breakthrough came when Juyeon employed the encoderdecoder framework, allowing the decoder to use MRI physics to regularize the target model. Integrating physics prior knowledge into the deep neural network improves generalization over previous methods and helps bridge the gap between theory and practice in medical imaging. “It’s hard to get a lot of medical training data for MRF, with ground truth, quantitative values,” she points out. “Usually, people train a model with a synthetic dataset and test it on a real dataset. That’s what we did also, and we improved generalization performance. We believe that by using our techniques, hopefully, it can have better results on real MRI data for patients to improve personalized treatment and medical diagnosis.” Looking ahead, Juyeon intends to extend her research by considering spatial information, which has the potential to improve it further. Additionally, she recognizes that her work on MRF is just one piece of the puzzle. She hopes to explore other aspects of the process and whether the whole process works for real-world cases. “MRF is so interesting because it can be really helpful for medical diagnosis,” she says. “I think this is so valuable. I’mfascinated by how we can include prior knowledge in deep neural networks, so it’s so interesting to put our welldeveloped physics information into the deep neural network.” Originally from Seoul in South Korea, Juyeon embarked on her academic journey at Cambridge two years ago when she started her PhD. “I like it so much,” she smiles. “There are a lot of other PhD students living together. Also, I live in the college, so it’s nice to interact with them and focus on the study.” 4 DAILY MICCAI Wednesday Poster Presentation
In a glimpse of life at Cambridge, she recalls a unique tradition: “In Cambridge, we have formals for each college, so every week, we wear formal clothes and gowns and have a really special dinner. I think this makes Cambridge special.” Reflecting on the interview, Juyeon describes it as a very comfortable experience before revealing that we sadly will not experience her presenting talents at MICCAI this week. However, the work will be ably represented by its second author, Pingfan Song. To learn more about this work, visit Poster 3 this morning at 09:3011:00 in the Poster Hall. 5 DAILY MICCAI Wednesday Juyeon Heo MICCAI Daily Publisher: RSIP Vision Copyright: RSIP Vision Editor: Ralph Anzarouth All rights reserved Unauthorized reproduction is strictly forbidden. Our editorial choices are fully independent from MICCAI, the MICCAI Society and MICCAI 2023 organizers.
Cortical thickness, the thickness or depth of a thin ribbon of gray matter surrounding the white matter in the cerebrum of the brain, has emerged as a potential biomarker for a range of neurodegenerative diseases and psychiatric conditions. One notable example is multiple sclerosis, where the rate at which the cortex thins provides essential information on whether a patient’s disease is being well controlled. There are several open-source tools available for quantifying cortical thickness. Although AI-based tools have started making their way into clinical settings for evaluating neurodegenerative diseases, they primarily focus on cortical volume rather than thickness. It’s a subtle yet significant distinction, as cortical volume comprises two key components: surface area and thickness. “These measures are typically used in large cohort studies,” Richard says. “For example, you can take a large cohort of epilepsy patients and identify that relative to matched healthy members of the population, particular regions of their brains tend to show a reduction in cortical thickness. These are often regions functionally linked to each other. We want to leverage these changes in thickness to determine at an earlier stage whether a patient will likely have one of these diseases.” While these tools are primarily used for research purposes today, it is hoped that, in the future, they could be applied more reliably to individual patients to aid in early diagnosis and personalized treatment. “These tools are not yet sufficiently accurate on an individual patient 6 DAILY MICCAI Wednesday Poster Presentation Richard McKinley is a Senior Researcher at Inselspital, the University Hospital of Bern. His paper on cortical thickness has been accepted as a poster, and he speaks to us ahead of his presentation this afternoon CortexMorph: fast cortical thickness estimation via diffeomorphic registration using VoxelMorph
7 DAILY MICCAI Wednesday Richard McKinley level,” he confirms. “We can use them as a research tool to see, in general, epilepsy patients will have this particular pattern of atrophy or one pattern of atrophy in frontotemporal dementia and another in Alzheimer’s. It allows us to study the different mechanisms by which these diseases occur.” Richard adds that if cortical thickness analysis were applied to patients, it would be one element of a comprehensive diagnostic process considering various factors, including neurological tests and clinical assessments. “These methods only give us one piece of the puzzle,” he asserts. “If that piece doesn’t fit with everything else, then you don’t diagnose a patient with a particular disease.” Existing commercial tools, like brain imaging software FreeSurfer, provide volumes of the different lobes of the brain as biomarkers to radiologists. Though valuable, these markers are less sensitive to diseases than cortical thickness. Also, they take a significant amount of time – upwards of 8-10 hours – to process data for a single patient, presenting a clear challenge as clinical workflows demand quicker and more precise solutions. “Our goal initially was to produce something able to estimate cortical thickness, which was as reliable as the existing tools,” Richard recalls. “Now, we have evidence to suggest we have something more sensitive and reproducible while also running in a matter of seconds!” He adopted an approach that is becoming increasingly important in the medical imaging field by reframing a problem previously solved using lengthy iterative algorithms. In doing this, he drew
inspiration fromVoxelMorphand its successor works, which took the problem of deformable image registration, formulated it as the loss function of a deep neural network, and then trained the neural network to solve it. “This, for me, is an excellent empirical but also elegant way to derive a solution to a problem,” he says. “You can have 20 pages of excellent theory explaining why your iterative algorithm works and converges well, but in the end, the proof of the pudding is in the validation. You don’t start from the principles of trying to solve your problem; you define what the parameters are for your problem to have been solved, and then, using gradient descent, you search for a solution.” Interestingly, Richard built the prototype for this method several years ago, but when you build a system for calculating cortical thickness, how do you validate it if there is no ground truth? Luckily, a paper by Rusak et al. last year had the answer: a synthetic phantom built using a GAN. “They generated 20 subjects with different levels of cortical thickness reduction and showed that a predecessor method to ours was very sensitive to these reductions,” he tells us. “It’s more sensitive than FreeSurfer. This work gave me the final piece of the puzzle. Now, I have a dataset coming from an independent group. It’s one thing to say my deep learning approach is close to the existing approach, but can it do the same job of resolving 8 DAILY MICCAI Wednesday Poster Presentation
9 DAILY MICCAI Wednesday these cortical thickness differences? Indeed, it does. In fact, it is even slightly more sensitive.” Did he get the chance to talk to Filip Rusak about his work? “Yes, a little bit,” he reveals. “After this paper was published, I was invited to be the examiner of his PhD thesis, which was very nice.” In terms of the next steps for this work, Richard says there are three key directions he hopes to pursue. First and foremost is validation and determining whether this method offers a similar or even superior ability to distinguish between different diseases compared to existing methods. Christian Rummell, the other author on this paper, has a currently funded project with the Swiss National Science Foundation to build an open-source suite of neuroimaging tools, allowing people to apply methods like CortexMorph in their own studies in their own hospitals. “We want to integrate these tools into this platform so people can use this method for free and, most importantly, easily,” he adds. “Before we do that, we need to validate Richard McKinley
10 DAILY MICCAI Wednesday Poster Presentation whether it really works –not just on healthy controls, but also on the ability to distinguish between diseases.” The folding of the human brain poses another significant challenge to surmount in accurately measuring cortical thickness. The brain sort of folds in on itself, and from the perspective of MRI, its surfaces virtually touch. From the segmentation, you effectively do not see a difference between the two banks of the sulcus, which present as a thick mass of gray matter. “FreeSurfer does a decent job of resolving these sulci, but often it doesn’t correctly go all the way down into the sulcus, and so you have an incorrect identification of cortical thickness,” Richard explains. “We believe, but have to validate, that our method gives a better resolution of these sulci. If it doesn’t, we have some ideas by going to super-resolution, for example, by leveraging high-field MRI to give us some basis for training our models to resolve these sulci better.” Finally, most other methods give a point estimate of cortical thickness, leaving clinicians unaware of the potential margin of error. Another next step is to produce a distribution of plausible cortical thickness values, or error bars, using ensemble methods, allowing a better understanding of measurement uncertainty and identifying regions of the brain where measurements may be less reliable. At Inselspital, Richard works in a group called the Support Center for Advanced Neuroimaging, a multidisciplinary research group with MDs, physicists, computer scientists, and psychologists interested in interpreting and quantifying imaging of the human brain. Before we finish, he is keen to mention and appreciate the work of his recently graduated PhD student, Michael Rebsamen. “Michael worked together with me on developing DL+DiReCT, which is the foundation of this work, funded by a grant from the Novartis Research Foundation,” he tells us. “Without that foundational work, we wouldn’t have been able to do this.” To learn more about Richard’s work, visit Poster 6 this afternoon at 14:30-16:00 in the Poster Hall. “We believe, but have to validate, that our method gives a better resolution
11 DAILY MICCAI Wednesday UKRAINE CORNER “I am Andriy Myronenko, a scientist at NVIDIA. My research is focused on deep learning for 3D medical image analysis. At MICCAI 2023, I participated and won 3 segmentation challenges. Our AI algorithm, based on MONAI, got 1st place in KiTS23, SEG.A23 and MVSEG23 challenges.”
In this paper, Luyi and joint first author Tianyu Zhang contend with two fundamental questions in multi-sequence MRI synthesis: how to quantify the contribution of different input sequences in synthesizing a missing sequence and how to estimate the quality of the generated image at the pixel level. Medical professionals rely on multiple MRI sequences for diagnosis and prognosis in many clinical scenarios. However, certain sequences are sometimes missing due to various factors and need to be synthesized. Recent works have primarily overlooked a crucial aspect: understanding which input sequences contribute more significantly to the synthesis process. Also, many works provide uncertainty maps presenting the uncertainty of output results and attention-based techniques that output attention maps to analyze regions of uncertainty in images. However, uncertainty-based methods require multi-time outputs to calculate standard deviation while attentional-based methods have limitations in terms of the low quality of the visualizations. “These two questions concern the synthesis model’s explainability and reliability,” Luyi tells us. “Unlike natural image analysis, our medical image analysis model must be more reliable and explainable because we need to use it in a clinical environment.” The researchers faced several challenges during their work. They used the BraTS2021 dataset, which includes four sequences: T1, T1Gd, T2, and Flair. There are many input combinations to synthesize a missing 12 DAILY MICCAI Wednesday Poster Presentation An Explainable Deep Framework: Towards Task-Specific Fusion for Multi-to-One MRI Synthesis Luyi Han is a PhD student at Radboud University Medical Centre and the Netherlands Cancer Institute. He speaks to us about his work on multi-sequence MRI synthesis ahead of his poster this afternoon.
sequence, and they had to create a single model capable of handling multiple input combinations and output scenarios. They also needed to quantify the contribution of each input sequence. “To solve this, we provide a taskspecific weighted average,” Luyi explains. “When we want to quantify the contribution, we think about the weighted average because if they contribute more, they have higher weight. We input the module with a combination of the models. For example, if we have T1, T2, and Flair and want to output T1Gd, we make a zero-one code to present this conditional input. For example, we can build it as 1011 and 0100 to present this input and output condition. With this condition, we can use a fully connected layer to predict a weight. This weight can be used as a weighted average for the sequence fusion. We can train the model and learn the weighted average based on the fully connected layer.” Also, Luyi and Tianyu designed a task-specific attention module. This module employs conditional channel and spatial attention to generate residual attentional fused features to refine the synthesis performance. The fusion feature is divided into a directly weighted average from the input sequence features and an attention model to generate residual fused features. “We use the weighted average feature as a basic feature,” Luyi continues. “We can directly reconstruct the image from this feature. Also, we build the attention 13 DAILY MICCAI Wednesday Luyi Han
14 DAILY MICCAI Wednesday Poster Presentation
15 DAILY MICCAI Wednesday Luyi Han module to generate residual attentional features. Then, we add the basic features with the attentional residual features together, and we can force it to reconstruct the target sequence. With these residual attentional features or without these attentional features, we can get two reconstructed images.” A task-specific enhanced map (TSEM) is proposed by subtracting these two reconstructed images. This map works like an attention map to identify which regions do not synthesize well in the image but at the pixel level. Unlike uncertainty, which requires multi-time outputs, it can be obtained in one inference. Looking to the future, Luyi tells us that the implications of this research extend beyond multisequence MRI synthesis. The model’s versatility allows it to be adapted for other tasks, such as segmentation or classification. “We also want to use this model on multi-sequence breast MRI,” Luyi reveals. “My PhD project is on using multi-sequence breast MRI to predict the response to neoadjuvant therapy, so I think this is the next step for this paper. My last paper was about longitudinal registration on breast MRI and is still under review. Another work is on 3D/4D multi-sequence MRI synthesis – not only on the breast but also on the brain. That’s also under review and is the basis of this paper.” To learn more about Luyi’s work, visit Poster 6 this afternoon at 14:30-16:00 in the Poster Hall.
16 DAILY MICCAI Wednesday Women in Science Read 100 FASCINATING interviews with Women in Science! Lisa Koch is a group leader for machine learning and medical diagnostics at the new Hertie Institute for AI in Brain Health at the University of Tübingen in Germany. She is also the Lead Organizer of the Domain Adaptation and Representation Transfer (DART) workshop, tomorrow at MICCAI 2023. Lisa, what brings you to MICCAI? Actually, the main reason I'm here this year is we are organizing DART, the workshop for learning domain adaptation and robust representation. Besides that, what is your work about? Very generally, maybe a little bit about my background. I've been doing medical image analysis and machine learning for medical applications for a very long time now. I started out in academia, and then after my first postdoc, I felt like I needed to leave academia to
17 DAILY MICCAI Wednesday feel more impact, and I joined a startup for a few years. You can name it if you want. Actually, the startup doesn't exist anymore. Okay, so you cannot name it… [laughs] I cannot name it. However, we were developing machine learning and data science approaches for women's health using physiological time series data. The company started out as more of a lifestyle product but turned into a medical device. During that time, I learned a lot about the actual realworld needs for AI in a high-risk situation. I decided to come back to research and found that these experiences in a medical device startup really motivated me to sharpen my vision for what I want to do. So, I'm very interested in trustworthy AI and safe applications. Really working with the clinicians to figure out what they need, how they interact with our algorithms, what problems we need to solve, and how. Is this what you always wanted to do? I haven't had a very clear path. Even before I started my PhD, I spent a year in industry. So I've been going back and forth a little bit and I think it's maybe a slightly unique trajectory, but I think it's good. I like where I'm at and I like that I have a pretty clear idea of why I'm doing the things I'm doing. Tell us, why are you doing the things you're doing? [hesitates for a moment] Well, I think from a technical perspective, what we are doing is very exciting. Now is a great time to be developing these methods. There's lots happening and I think the field we are in, we can make a real difference. Especially now that there are so many applications that are reaching real-world human-level performance. I think it's a very important time to figure out how we can actually make the transition. How can we bring this to clinical reality and to actual patient benefit? Is your drive patient benefit? Yes, ultimately. I mean, patient benefit of whoever is using the technology. Patient benefit in the end, of course. Tell us about this Dart workshop. What are you guys going to show? Actually, we are going to have a lot Lisa Koch
of papers presented in a very short amount of time. We will do our best to use the time well. We had a record number of submissions this year, I guess similar to MICCAI itself. I think we had 35 submissions. We accepted about half of them. So, we will have 15 or 16 papers presented on different topics around robust and intraverbal presentations. So, a lot about domain adaptation, domain generalization, and harmonization. Also, fusion learning and continual learning. What do you expect the main takehome thoughts to be? I think these workshops are great because people come there with similar research interests. We try to make a program that is interactive, that allows the researchers to actually get some feedback from the community. We have too many papers to give everybody an oral platform. So, we have a poster session in which we will proceed with a spotlight for everybody. In this way, we can really draw the audience to the posters as well. So we just hope that everybody gets as much community interaction as possible. Was that your idea? Oh, no. The workshop is now in its fifth year. It was started by Kostas Kamnitsas, and it has been going really well since the start, basically. I joined the organizing committee two years ago and this year, I'm the lead organizer. That is very impressive! What will you become next year? So next year we will see. [smiles] We will see who is on board. I think it's a nice event that I want to stay partof. Let’s speak a little bit about your future. What are your wildest dreams? [hesitates for a moment] I want to continue what we are doing at the moment. In my group, I said I'm interested in trustworthy AI and actually making AI useful in reality. We are working on different aspects of that. We work a lot on interpretable methods, so inherently interpretable methods. 18 DAILY MICCAI Wednesday Women in Science
19 DAILY MICCAI Wednesday Lisa Koch … failure is at the very essence of research because we are always looking into what we don't know yet!
20 DAILY MICCAI Wednesday Women in Science Also, we use counterfactual explanations as explanations that we can give to clinicians to reason about our machine learning models. We work quite a bit on disease progression modeling, population modeling, understanding our image data, and image distributions. In terms of applications, we are working very closely with the Eye Hospital in Tübingen so we work on ophthalmic images a lot. One of the reasons for this is that we have these very close collaborations in place, we have access to the doctors. Doctors are interested in doing research with us. What has been happening in the past few years and is coming to fruition now is that we have worked on several fronts that are really coming to a clinical study state, where we can study the effect of using our models in interaction with the clinicians. I think this is a very important direction for us that we want to continue. We are seeing the first results now. And in the first results, we found that some methods, when we put them into clinicians actually didn't give us any benefit. That's important to know. But we also saw some first results that looked very exciting! I think we will have some results to show soon. Lisa, why do you think doctors cooperate with you? What brings them to you? Well, I think some of the clinicians that we work with have the capacity to do research on the site. They have the time. In Tübingen, we have a very nice situation. We have a very strong hospital and we have a very strong AI research community. I find that the atmosphere is an atmosphere of collaboration. This is probably not unique, but it's a very pleasant way to do research and to interact. Our building is also next to the hospital. So we can go have lunch together or have coffee together. From being so close to doctors, what was the one thing that inspired you most? What did you learn from a doctor that was extremely precious for you? Well, it’s not just one thing. It’s very important for me to understand how they spend their days. So, we organized shadowing hospital visits with the doctor. Myself and the students would follow them around for a day to see how they interact with the patient and how they interact with imaging. This helps us
21 DAILY MICCAI Wednesday Lisa Koch to figure out from observation and from talking to them where AI can be useful. Also, to discover where imaging is used, as opposed to maybe just talking to the patients or other variables. Tell me more about Tübingen. Why Tübingen? What is special about it that we don't know? Tübingen is a small town in the south of Germany. It used to be famous mostly for theology. It's a very old university town, but nowadays it's famous for its AI research. So it's a great place to be! The Max Planck Institute for Intelligence Systems is there, right? It is! And around this, there is a growing community and just excellent infrastructure and excellent fundamental machine learning research. With the clinic, and it's a very strong clinic, it's the perfect place to do the kind of work that we do. Are you going to do research all your life? [hesitates for a moment] I hope so. I do think my interests are varied. So, I think I can be happy in different settings. What I like about academic research is the content of my work, including the topic area. But I also like the job description of an academic researcher. It's so diverse. I love working with the students, teaching, the management and supervision skills, and the technical progress that we make. Lisa, you look very upbeat. I can’t find any weak points where you say, “I am unhappy with this. Maybe I am not doing the right thing.” Is everything perfect? What could go wrong? I mean, it's a very challenging job that we have, with lots of failures. I think failure is at the very essence of research because we are always looking into what we don't know yet! I think if you want to be happy doing research, you have to accept that, somehow. And yeah, I think I have come to terms with that. [laughs] Don’t get me wrong, I love interviewing upbeat people! I just want to learn their secret, that's all. [both laugh] So what's your secret? How do you stay so positive? [hesitates for a moment] I have struggled in the past and I still struggle with some aspects of our work and our job. But I think I've gotten a lot of perspective on what's important for me personally and in my job. I think also through these experiences outside of academia, but
22 DAILY MICCAI Wednesday Women in Science also through family, I realized that there's more than just the next paper. True! I interviewed Jessica Sieren a few years back and she told me if you are doing a PhD and you don't have a major crisis, you are not doing it right… [laughs] Yeah, I guess that's probably correct. So, what was your biggest crisis? I mean, during my PhD, there were phases where the projects were not going so well. At some point, for me, at least, I noticed that my personal happiness was correlated with the results of my experiments. I also realized that we attach a lot of personal sense of worth to those experiments. I think you have to decouple that, though. Your final words. I look forward to welcoming people to our workshop on Thursday!
Do you enjoy MICCAI 2023? Do you enjoy reading our dailies? MICCAI 2023 does not end this week! Like every year, Computer Vision News will publish the BEST OF MICCAI in the issue of November. Yes, in just 3 weeks! GET THE BEST OF MICCAI! Subscribe for free and get the BEST OF MICCAI in your mailbox. Computer vision News. Meet the scientist. 23 DAILY MICCAI Wednesday BEST OF MICCAI 2023
24 DAILY MICCAI Wednesday Workshop Preview In recent years, much discussion and research funding has been invested in early cancer detection. The Cancer Prevention through early detecTion (CaPTion) workshop aims to bring this topic to the forefront of theMICCAI community. Before the workshop’s first edition last year, organizers noted a shift in what they perceived as a very technical conference, with MICCAI seeing a growing focus on interaction with clinicians and even introducing a clinical day. “Early detection is not only detection; there is a strong imaging component, and this fits very well with MICCAI,” Bartek explains. “There is lots of preclinical research which uses microscopy and all of the variants of preclinical imaging. They process and analyze images, not necessarily to detect cancer, but to understand the mechanism of how cancer develops and progresses. We wanted to bring this to the MICCAI community and focus people around the application rather than technology.” The CaPTion workshop is critical to this mission. Revolving entirely around early cancer detection, it covers various facets of the application, from biology and screening to integration. It brings together clinicians, researchers, and Bartek Papiez is an Associate Professor at the Big Data Institute in Oxford, leading the Medical Image Analysis and Machine Learning group. He is co-organizing an innovative MICCAI workshop on early cancer detection and speaks to us ahead of tomorrow’s (Thursday) main event in Vancouver. Cancer Prevention through early detecTion (CaPTion) @ MICCAI2023 Workshop Ziang Xu et al.
25 DAILY MICCAI Wednesday CaPTion- Early Cancer Detection industry players, emphasizing the potential for translating groundbreaking research into realworld solutions. “We wanted kind of a pitch to all of the MICCAI people – maybe you already have interesting technology which can be applied to early cancer detection,” Bartek says. “Just come, show us, and show the other people who might be interested. This may be a small change, but the new technology can make very rapid developments in actual detection!” A crucial aspect of this workshop is the synergy between the right people and the right datasets. Creating data pools or lakes has historically been pivotal in research and development. These resources are precious for smaller labs that might lack the capacity to gather extensive datasets independently. By fostering interactions and networking through these datasets, the ambition is to expedite progress in the field. Ultimately, this collaborative push holds the promise of translating advancements into tangible benefits for patient healthcare. With CaPTion scheduled for the last day of MICCAI, organizers are keen to encourage conference attendees to put off their sightseeing for one more afternoon and come to what promises to be a fascinating event. But what makes CaPTion stand out from the other workshops, all vying for attention simultaneously? “There are multiple reasons,” Bartek reveals. “We’ve already completed the paper selection process, and I’m very excited that we’ve selected a number of very, very good quality papers, which I believe will be of interest to everyone. Come and hear about the new methods developed, the new datasets collected, and the new clinical evaluations conducted.”
26 DAILY MICCAI Wednesday Workshop Preview As well as being a gateway to groundbreaking research, the workshop promises a unique networking opportunity and the chance to continue building a vibrant community. It has a multidirectional approach, and organizers aim to bridge the gap between medical imaging and clinical science with a diverse lineup of distinguished keynote speakers at the forefront of their fields. Sravanthi Parasa, Anne Martel, and Sir Michael Brady will deliver enlightening talks covering technology and methodology core to MICCAI’s mission, including a personal journey of translating technology from the university lab to real-world patient care via an innovative spinout. “I think that’s going to inspire people to pursue such research – pursue more like a commercial side of the research, basically,” Bartek tells us. “It’s to emphasize that it’s possible, it’s doable, and it can make a real impact to the patients!” Bartek’s group at the Big Data Institute has around 12 people working on different biomedical imaging problems. It works with clinical imaging and large population studies and supports preclinical scientists and other research groups with image analysis. For Bartek, the ideal impact of CaPTion this year and in future years transcends academic achievement. Instead, it is the realization of technologies and methodologies presented at the workshop being picked up, translated to the clinic, and going on tosavepeople’s lives. “It may sound a little weird when a researcher says that the main impact is not only the paper, but the long-term patient benefit,” Bartek points out. “A paper describing a certain mechanism for how cancer progresses would also be an excellent outcome because we can learn from it and use it to develop a better screening methodology. The big dream is that one of the papers is a technology that will improve the detection or understanding of cancer. It’ll make a real impact on those who are the most important here: the patients.” In the short term, CaPTion is about uniting a community with a common goal: revolutionizing early cancer detection. The workshop aims to attract fresh minds, encourage collaboration over competition, and lay the groundwork for an enduring movement. The hope is that this collective spirit will spur more research, new technologies, and ultimately a more significant impact on the lives of those affected by cancer. “It’s possible, it’s doable, and it can make a real impact to the patients!”
27 DAILY MICCAI Wednesday CaPTion- Early Cancer Detection “It’s the same for MICCAI,” Bartek adds. “We hope there will be more and more papers submitted to the main conference on early cancer detection because if people become more interested, more work will be done. It’s like building a backbone, building infrastructure, building a movement so that people want to work and invest time in that area.” The workshop will be giving awards for Best Paper and Best Poster. Once again, it has funding from Satisfai Health, whose CEO and founder was a keynote speaker last year. Satisfai will also participate in the workshop, demonstrating a solid orientation toward translation. “I‘d like to invite everyone to participate in the workshop,” Bartek declares. “I would love to see as many people as possible, despite the fact we’re on the very last day of MICCAI. Stay one more day in Canada and visit beautiful Vancouver, but before you leave the main MICCAI conference, please come and join our workshop and hear the exciting keynote speakers. We hope it will be another successful event for the CaPTion community, the MICCAI community, and the patients!” Ziang Xu et al. The Cancer Prevention through early detecTion (CaPTion) workshop will take place on October 12 (tomorrow) at 13:30.
28 DAILY MICCAI Wednesday Keynote Speaker
29 DAILY MICCAI Wednesday Yann LeCun Did you read Yann’s interview? It is fascinating ☺
RkJQdWJsaXNoZXIy NTc3NzU=