Our paper, Endoscopic navigation in the absence of CT imaging, was presented at MICCAI in Granada, Spain! The talk was presented during the CAI session in the García Lorca Hall and a poster presentation followed the oral session.
PhD student Xingtong Liu presented our paper, Self-supervised Learning for Dense Depth Estimation in Monocular Endoscopy, at the CARE (Computer Assisted and Robotic Endoscopy) Workshop at MICCAI in Granada, Spain. This paper was awarded the best paper award at the workshop and the presentation was voted the best presentation.
Our MICCAI paper Endoscopic navigation in the absence of CT imaging is selected for oral presentation in Granada, Spain! It also earned a travel award based on reviewers' scores. We will update this space with information about the day and time of the presentation as we learn more! Slides wil be available on this website after the talk. Update The talk will be presented on the 19th of September at 9:45am in the García Lorca Hall.
Codebase with implementations of most-likely point and deformable most-likely point algorithms is now available on github. This includes code for our MICCAI paper Endoscopic navigation in the absence of CT imaging. An extensive README will walk you through the steps to install dependencies and our codebase as well as steps to run a test script to ensure everything works as expected!
Our paper, "Self-supervised Learning for Dense Depth Estimation in Monocular Endoscopy", by Xingtong Liu was accepted at MICCAI workshop on Computer Assisted Robotic Endoscopy (CARE). This paper explores the use of structure from motion and depth consistency among frames to predict dense depth maps from single frames in sinus endoscopy. Preprint is available!
I will attend Rising Stars in Biomedical, a two-day career development workshop, after being nominated by a montor and selected from a pool of outstanding candidates. The workshop, organized by JHU and MIT faculty, aims to connect top junior researchers in the biomedical field. I will get a chance to present my work to other participants and faculty and also to network with peers in both academia and industry.
Preprint of our paper, Endoscopic navigation in the absence of CT imaging, is now available on arXiv. Read to find out how navigation during minimally invasive procedures is possible even when corresponding patient CT scans are not available.
Our paper, "Endoscopic navigation in the absence of CT imaging", was accepted at MICCAI. This paper presents a method that enables navigation in the absence of patient specific CTs by registering video features to a statistical shape model built from population CTs. This paper also computes statistical confidence in each registration allowing clinicians to know when the system can be trusted.
Preprint of our paper, Evaluation and Stability Analysis of Video-Based Navigation System for Functional Endoscopic Sinus Surgery on In-Vivo Clinical Data, is now available in the Early Access section of IEEE TMI. Read to learn about video-CT registration with stability analysis for endoscopic procedures!
Our paper, "Evaluation and Stability Analysis of Video-Based Navigation System for Functional Endoscopic Sinus Surgery on In-Vivo Clinical Data", was accepted for publication by IEEE TMI. This paper presents a pipeline for rigid video-CT registration that produces submillimeter registrations and also performs a stability analysis for the registration. This allows a clinician to gauge the reliability of the computed registration. Watch this space for the published paper!
My thesis, "Deformable registration using shape statistics with applications in sinus surgery", was accepted by the library at the Johns Hopkins University. The thesis will be published at the end of the academic year, and I will be hooded at the graduation ceremony in May. The thesis will be available here as soon as it is published. Hold tight!
Successfully defended my dissertation! My talk presented some of our latest work on deformable registration using statistical shape models. The talk was attended by my thesis committee consisting of Russ Taylor, Greg Hager, Austin Reiter and Masaru Ishii, as well as by colleagues and friends. I had a great time sharing my work with everyone, and hope to continue to do so. Code from my registration framework will be available soon, as well as slides from my defense. Watch this space for more!
Named one of 7 Provost's Postdoctoral Fellows for the year 2018. This is a great opportunity for me to continue my work at the Johns Hopkins University once I have completed my Ph.D. This fellowship is part of the Faculty Diversity Initiative, which is one of many programs outlined in the JHU Roadmap on Diversity and Inclusion. All fellows receive ne year of salary at the current NIH level of support to continue work in academia. For more information, visit the program page. The fellows named can be found here.
Two new papers on registration techniques have been submitted to two different journals for review by our team. One of the papers is a rigid registration paper led by Simon Leonard, while the other is my effort towards a deformable registration framework. Watch this space for updates as the papers are published and we are able to share more!
The Department of Computer Science at the Johns Hopkins University released a PhD student profile page highlighting student research areas and publications allowing people to learn more about the variety of work being done here. A recent PhD graduates profile page is also available to learn where Johns Hopkins CS PhD students go after graduation!
Volunteered with Johns Hopkins Engineering Innovation program to help with an activity designed by Anand Malpani to teach high school students about the role of technology in surgical education. The students came from all around the world, and learned about the various robotic skills required in open and laproscopic surgery.
Video of my presentation of our paper simultaneous segmentation and correspondence improvement using statistical modes at SPIE Medical Imaging conference earlier this year is out now! This paper was selected as one of the Robert F. Wagner best student paper finalists, and presents an elegant method to simultaneously improve segmentation and correspondence between homologous meshes.
Was inducted into the International Honor Society for the Computing and Information Disciplines, Upsilon Pi Epsilon (UPE). UPE is the first honor society dedicated to the discipline of computer information systems and computer science.
Our paper, simultaneous segmentation and correspondence improvement using statistical modes, Robert F. Wagner best student paper finalist, is out now and available for download from the SPIE Digital Library. In this paper, we present a method that iteratively improves both segmentation as well as correspondence between segmentations by using statistical shape models (SSMs) not only to improve correspondence, but also to constrain the movement of vertices during segmentation improvement. To learn more, take a look at the publication! If you do not have a subscription to the SPIE Digital Library, feel free to contact me for a copy of the paper!
Our paper, Simultaneous segmentation and correspondence improvement using statistical modes, which was accepted to SPIE Medical Imaging conference, is a Robert F. Wagner best student paper finalist. All finalists will be announced and recognised at the start of the Plenary and Awards session on Monday, 13th February, at 4pm in Orlando, FL. The runner-up and winner will be announced after this! Update The winner and runner-up for the Robert F. Wagner best student paper award were announced at the conference on Monday, the 13th of February. More inforamtion and updates can be found here.
Master's project manuscript is now public! In this paper, we present a new method for computing approximate geodesic distances. We introduce the wave method for approximating geodesic distances from a point on a manifold mesh. Our method involves the solution of two linear systems of equations. One system of equations is solved repeatedly to propagate the wave on the entire mesh, and one system is solved once after wave propagation is complete in order to compute the approximate geodesic distances up to an additive constant. However, these systems need to be pre-factored only once, and can be solved efficiently at each iteration. All of our tests required approximately between 300 and 400 iterations, which were completed in a few seconds. Therefore, this method can approximate geodesic distances quickly, and the approximation is highly accurate. Watch videos about the project on the project page!
Served on the panel of the GRACE Mentoring Dinner along with Prof. Joanne Selinksi, Prof. Muyinatu Bell, Dr. Louise Sengupta, and undergraduate CS student, Paige Senal. The event, organized by GRACE in Clipper Room, Shriver Hall on the 5th of December, was attended by over 90 people. We discussed our experiences in our fields, as well as issues faced by women in engineering and how we can combat them. You can read more about the event here.
Presented our paper Anatomically Constrained Video-CT Registration via the V-IMLOP Algorithm at MICCAI in Athens, Greece. In this paper, we present an algorithm developed to aid surgeons during functional endoscopic sinus surgery (FESS). FESS is a surgical procedure used to treat acute cases of sinusitis and other sinus diseases. FESS is fast becoming the preferred choice of treatment due to its minimally invasive nature. However, due to the limited field of view of the endoscope, surgeons rely on navigation systems to guide them within the nasal cavity. State of the art navigation systems report registration accuracy of over 1mm, which is large compared to the size of the nasal airways. We present an anatomically constrained video-CT registration algorithm that incorporates multiple video features. Our algorithm is robust in the presence of outliers. We also test our algorithm on simulated and in-vivo data, and test its accuracy against degrading initializations. The manuscript is also available on arxiv, and the code is coming soon. Read more on the project page!
Our paper, Simultaneous segmentation and correspondence improvement using statistical modes, has been accepted for an oral presentation at SPIE Medical Imaging conference in Orlando, FL, from the 12th-16th of February, 2017. The talk will be held on the 14th of February. Please join!