Transparent Blue by BryantSmith.com

About Me

Welcome to my homepage! My name is Sheng You. I was born and grew up in China. I moved to the United States since 2007 for graduate school. I am currently a Ph.D. student in Electrical Engineering at Northeastern University. I work at Cognitive Systems Lab (CSL) under the advise of Pro. Deniz Erdogmus. My research interests are biomedical image processing, computer vision, machine learning, signal detection and estimation. Now I am also working full time as an image analysis scientist at Presage Biosciences.

 

Research     back to top

Microvascular blood flow analysis in sublingual surface videos

Studying the microcirculation of blood in vessels, especially capillaries and small blood vessels plays an important role in monitoring human physiological health. Monitoring the microcirculation can assist physicians in diagnosis of diseases such as sepsis, diabetes, and hypertension. Moreover, quantitatively analyzing red blood cell (RBC) velocities and vessel geometry can help to detect and diagnose these pathological conditions. Vessel diameter, the proportion of perfused vessels (PPV), perfused vessel density (PVD), functional capillary density (FCD), microvascular flow index (MFI) are all important features to quantitatively evaluate the microcirculation.

 

Estimation of RBC velocities is challenging. Even though some imaging techniques, such as the orthogonal polarization spectral (OPS), and the sidestream dark field (SDF) can provide high quality images of the microvasculature, vessels still have low contrast to the background. The presence of RBCs depend on the intensity change of each pixel over time. However, images are noisy, and illuminance of the images varies over space and time. Moreover, the camera may not be stable during the microcirculation videos. These image artifacts make it difficult to track RBCs and estimate their velocities. Typically, physicians manually trace small blood vessels and visually estimate RBC velocities. The task is labor intensive, tedious, and time-consuming. Therefore, our goal for this project is to develop a functional and fully automatic framework to estimate red blood cell velocites and extract vascular properties to assist the decision and diagnosis of human pathological conditions.

 

 

Segmentation of organs and tumors in 3D/4D CT for radiation treatment planning

Radiation therapy is one of the most effective options used in the treatment of about half of all people with cancer. A critical goal in radiation therapy is to deliever optimal radiation doeses to the observed tumor while sparing the surrounding healthy tissues. Radiation oncologists typically manually delineate normal and diseased structures on 3D/4D-CT scans.

 

I have worked on projects segmenting organs and tumors automatically in 3D or 4D CT scans using principal curve/surface algorithms. For segmentation in 4D-CT, the kernel density estimate based principal surface algorithm is applied. It incorporates the user inputs from the reference phase as an initialization and edge information from all the target phases. The initialized 3D contours are converged to the ridges of the edge maps by applying the principal surface algorithm and propagated consecutively and recursively through all the phases. For segmentation in 3D-CT, we present a semi-supervised method to segment the contours of organs represented by piecewise linear segments connected with a small number of points give the user's input in one or more slices as an approximate initialization. A few slices delineated by the users are selected as the reference slices. Delineations of these slices are treated as the initial contours and propagated to the ridges of the edge maps. These projected contours on the given slices are then propagated to slices above and below the given slices. Contour points are then downsampled to a small number of points while with the shape accuracy preserved based on a principal curve score.

 

 

Retinal vessel segmentation and feature analysis on the diagnosis of retinopathy of prematurity

The analysis and detection of retinal vessels is critical to diagnosing and detecting retinal diseases, such as diabetic retinopathy, retinopathy of prematurity(ROP), hypertension. Differences in retinal vessel diameters, lengths, tortuosity, reflectivity, branch bifurcations, and angles are important measurements underly the retinal phenotypes of these diseases. The isotropic Gaussian kernel Frangi filter is used to enhance the retinal vessels and measure the diameters of them. A multiscale principal curve projection and tracing algorithm is then applied to identify the centerlines of the vessels in the output image of the Frangi filter using the underlying kernel smoothing interpolation of the intensities.

 

ROP is a disease affecting low-birth weight infants. This disease is called plus disease which is characterized by tortuosity of the arteries and dilation of the veins in the posterior retina. Human diagnosis is often subjective and qualitative. Computer-aided image analysis has been used to improve the plus disease diagonosis in ROP by quantifying vascular features such as dilation and tortuosity. However, analyzing the variability of expert decisions and the relationship between the expert diagnosis and features, such as acceleration, tortuosity index, diameter, branching factor. This is an important gap in knowledge that will help improve clinical diagnosis, as well as optimize algorithms for computer-aided diagnosis. The analysis is based on mutual information and kernel density estimation on features.

 

 

 

Lung tumor tracking in real time for radiation therapy

Radiotherapy is an effective treatment technique for lung cancer. However, the movement of lung tumors during normal respiration makes it difficult to accurately irradiate the tumor. Precise lung tumor localization is vital to efficiently treating the tumor and avoiding unnecessary radiation exposure of normal tissues. Estimating the motion model of the tumor may lead to improved treatment planning and dose calculation throughout the therapy. 4D Computed Tomography (CT) images taken prior to treatment (to develop the patient's treatment plan) provide valuable information about the movement of the thoracic organs and the tumor. For each radiotherapy treatment session, a set of kilovoltage X-ray images (kV images) are acquired to aid in the alignment of the treatment target relative to the radiation beam. This project focuses on real time lung tumor tracking by incorporating information and labels from 4DCT on kV x-ray videos for treatment-day specific tumor motion models.

 

 

Publications:     back to top

Journals

[1] K. Keck, J. Kalpathy-Cramer, E. Ataer-Cansizoglu, S. You, D. Erdogmus, M. Chiang, Plus disease diagnosis in retinopathy of prematurity: vascular tortuosity as a function of distance from optic disc, The Journal of Retinal and Vitreous Diseases

[2] S. You, M. Massey, N. Shaprio, D. Erdogmus, Automatic red blood cell velocity estimation in sublingual microcirculatory videos, in prepration

[3] S. You, M. Massey, N. Shaprio, D. Erdogmus, Classification of pathological conditions based on the features extracted from the sublingual microcirculation, in prepration

 

Conferences

[1] S. You, M. Massey, N. Shaprio, D. Erdogmus, A Novel Line Detection Method on Space-Time Images for Microvascular Blood Flow Analysis in Sublingual Microcirculatory Videos, 2013 International Symposium on Biomedical Imaging (ISBI), San Francisco, USA 2013

[2] S. You, E. Ataer-Cansigolue, M. Massey, N. Shaprio, D. Erdogmus, Microvascular blood flow estimation in sublingual microcirculation videos based on a principal curve tracing algorithm, 2012 IEEE International Workshop on Machine Learning for Signal Processing (MSLP), Santander, Spain 2012

[3] E. Ataer-Cansizoglu, S. You, J. Kalpathy-Cramer, D. Erdogmsu, Observer and feature analysis on diagnosis of retinopathy of prematurity, 2012 IEEE International Workshop on Machine Learning for Signal Processing (MSLP), Santander, Spain 2012

[4] K. Keck, J. Kalpathy-Cramer, E. Ataer-Cansizoglu, S. You, D. Erdogmus, M. Chiang, Plus disease diagnosis in retinopathy of prematurity: vascular tortuosity as a function of distance from optic disc center, The Association for Research in Vision and Ophthalmology (ARVO), Fort Lauderdale, USA, 2012

[5] S. You, E. Bas, J. Kalpathy-Cramer, D. Erdogmus, Principal curve based retinal vessel segmentation towards diagnosis of retinal diseases, IEEE Conference on Healthcare Informatics, Imaging, and Systems Biology (HISB), San Jose, USA, 2011

[6] S. You, E. Bas, E. Ataer-Cansizoglu, J. Kalpathy-Cramer, D. Erdogmus, Principal curve based semi-automatic segmentation of organs in 3D-CT, 33rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Boston, USA, 2011

[7] S. You, E. Bas, D. Erdogmus, Extraction of samples from airway and vessel trees in 3D lung CT based on a multi-scale principal curve tracing algorithm, 33rd Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Boston, USA, 2011

[8] S. You, E. Ataer-Cansizoglu, J. Tanyi, J. Kalpathy-Cramer, D. Erdogmus, A novel application of principal surfaces to segmentation in 4D-CT for radiation treatment planning, IEEE International Conference on Machine Learning and Applications (ICMLA), Washington DC, USA, 2010

[9] E. Ataer-Cansizoglu, E. Bas, M. Ali Yousuf, S. You, W. D. D'Souza, D. Erdogmus, Towards Respiration Management In Radiation Treatment Of Lung Tumors: Transferring Regions Of Interest From Planning CT To Kilovoltage X-Ray Images, 32nd International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Buenos Aires, Argentina, 2010

 

 

Resume     back to top

Please find my CV here.

 

 

Contact     back to top

Mailing Address: 409 Dana Research Center, Department of Electrical and Computer Engineering, Northeastern University, 360 Huntington Avenue, Boston, MA 02115

 

Phone: +1-617-413-5503

 

Email: syou@ece.neu.edu