skip to main content
US FlagAn official website of the United States government
dot gov icon
Official websites use .gov
A .gov website belongs to an official government organization in the United States.
https lock icon
Secure .gov websites use HTTPS
A lock ( lock ) or https:// means you've safely connected to the .gov website. Share sensitive information only on official, secure websites.

Explore Research Products in the PAR
It may take a few hours for recently added research products to appear in PAR search results.


  • NSF Public Access
  • Search Results
  • Classification of Pancreatic Cysts in Computed Tomography Images Using a Random Forest and Convolutional Neural Network Ensemble
Title: Classification of Pancreatic Cysts in Computed Tomography Images Using a Random Forest and Convolutional Neural Network Ensemble
There are many different types of pancreatic cysts. These range from completely benign to malignant, and identifying the exact cyst type can be challenging in clinical practice. This work describes an automatic classification algorithm that classifies the four most common types of pancreatic cysts using computed tomography images. The proposed approach utilizes the general demographic information about a patient as well as the imaging appearance of the cyst. It is based on a Bayesian combination of the random forest classifier, which learns subclass-specific demographic, intensity, and shape features, and a new convolutional neural network that relies on the fine texture information. Quantitative assessment of the proposed method was performed using a 10-fold cross validation on 134 patients and reported a classification accuracy of 83.6%.  more » « less
Award ID(s):
1650499 1069147
PAR ID:
10054446
Author(s) / Creator(s):
Date Published:
Journal Name:
Medical Image Computing and Computer-Assisted Intervention − MICCAI 2017
Volume:
09/04/2017
Page Range / eLocation ID:
150-158
Format(s):
Medium: X
Sponsoring Org:
National Science Foundation
More Like this
  1. ; ; (, Scientific Reports)
    null (Ed.)
    Abstract Soybeans are an important crop for global food security. Every year, soybean yields are reduced by numerous soybean diseases, particularly the soybean cyst nematode (SCN). It is difficult to visually identify the presence of SCN in the field, let alone its population densities or numbers, as there are no obvious aboveground disease symptoms. The only definitive way to assess SCN population densities is to directly extract the SCN cysts from soil and then extract the eggs from cysts and count them. Extraction is typically conducted in commercial soil analysis laboratories and university plant diagnostic clinics and involves repeated steps of sieving, washing, collecting, grinding, and cleaning. Here we present a robotic instrument to reproduce and automate the functions of the conventional methods to extract nematode cysts from soil and subsequently extract eggs from the recovered nematode cysts. We incorporated mechanisms to actuate the stage system, manipulate positions of individual sieves using the gripper, recover cysts and cyst-sized objects from soil suspended in water, and grind the cysts to release their eggs. All system functions are controlled and operated by a touchscreen interface software. The performance of the robotic instrument is evaluated using soil samples infested with SCN from two farms at different locations and results were comparable to the conventional technique. Our new technology brings the benefits of automation to SCN soil diagnostics, a step towards long-term integrated pest management of this serious soybean pest. 
    more » « less
  2. ; (, 2019 IEEE International Ultrasonics Symposium (IUS))
    We evaluated training deep neural network (DNN) beamformers for the task of high contrast imaging in the presence of reverberation clutter. Training data was generated using simulated hypoechoic cysts and a pseudo nonlinear method for generating reverberation clutter. Performance was compared to standard delay-and-sum (DAS) beamforming on simulated hypoechoic cysts having a different size. For a hypoechoic cyst in the presence of reverberation clutter, when the intrinsic contrast ratio (CR) was -10 dB and -20 dB, the measured CR for DAS beamforming was -9.2±0.8 dB and -14.3±0.5 dB, respectively, and the measured CR for DNNs was -10.7±1.4 dB and -20.0±1.0 dB, respectively. For a hypoechoic cyst with -20 dB intrinsic CR, the contrast-to-noise ratio (CNR) was 3.4±0.3 dB and 4.3±0.3 dB for DAS and DNN beamforming, respectively. These results show that DNN beamforming was able to extend contrast ratio dynamic range (CRDR) by about 10 dB while also improving CNR. 
    more » « less
  3. ; ; ; ; ; (, Integrative Biology)
    Abstract The human embryo is a complex structure that emerges and develops as a result of cell-level decisions guided by both intrinsic genetic programs and cell–cell interactions. Given limited accessibility and associated ethical constraints of human embryonic tissue samples, researchers have turned to the use of human stem cells to generate embryo models to study specific embryogenic developmental steps. However, to study complex self-organizing developmental events using embryo models, there is a need for computational and imaging tools for detailed characterization of cell-level dynamics at the single cell level. In this work, we obtained live cell imaging data from a human pluripotent stem cell (hPSC)-based epiblast model that can recapitulate the lumenal epiblast cyst formation soon after implantation of the human blastocyst. By processing imaging data with a Python pipeline that incorporates both cell tracking and event recognition with the use of a CNN-LSTM machine learning model, we obtained detailed temporal information of changes in cell state and neighborhood during the dynamic growth and morphogenesis of lumenal hPSC cysts. The use of this tool combined with reporter lines for cell types of interest will drive future mechanistic studies of hPSC fate specification in embryo models and will advance our understanding of how cell-level decisions lead to global organization and emergent phenomena. Insight, innovation, integration: Human pluripotent stem cells (hPSCs) have been successfully used to model and understand cellular events that take place during human embryogenesis. Understanding how cell–cell and cell–environment interactions guide cell actions within a hPSC-based embryo model is a key step in elucidating the mechanisms driving system-level embryonic patterning and growth. In this work, we present a robust video analysis pipeline that incorporates the use of machine learning methods to fully characterize the process of hPSC self-organization into lumenal cysts to mimic the lumenal epiblast cyst formation soon after implantation of the human blastocyst. This pipeline will be a useful tool for understanding cellular mechanisms underlying key embryogenic events in embryo models. 
    more » « less
  4. ; ; ; ; ; ; ; (, Science Advances)
    Despite its importance in central nervous system development, development of the human neural tube (NT) remains poorly understood, given the challenges of studying human embryos, and the developmental divergence between humans and animal models. We report a human NT development model, in which NT-like tissues, neuroepithelial (NE) cysts, are generated in a bioengineered neurogenic environment through self-organization of human pluripotent stem cells (hPSCs). NE cysts correspond to the neural plate in the dorsal ectoderm and have a default dorsal identity. Dorsal-ventral (DV) patterning of NE cysts is achieved using retinoic acid and/or sonic hedgehog and features sequential emergence of the ventral floor plate, P3, and pMN domains in discrete, adjacent regions and a dorsal territory progressively restricted to the opposite dorsal pole. This hPSC-based, DV patterned NE cyst system will be useful for understanding the self-organizing principles that guide NT patterning and for investigations of neural development and neural disease. 
    more » « less
  5. ; ; ; (, Integrative Biology)
    Abstract Throughout the body, epithelial tissues contain curved features (e.g. cysts, ducts and crypts) that influence cell behaviors. These structures have varied curvature, with flat structures having zero curvature and structures such as crypts having large curvature. In the ovary, cortical inclusion cysts (CICs) of varying curvatures are found, and fallopian tube epithelial (FTE) cells have been found trapped within these cysts. FTE are the precursor for ovarian cancer, and the CIC niche has been proposed to play a role in ovarian cancer progression. We hypothesized that variations in ovarian CIC curvature that occur during cyst resolution impact the ability of trapped FTE cells to invade into the surrounding stroma. Using a lumen model in collagen gels, we determined that increased curvature resulted in more invasions of mouse FTE cells. To isolate curvature as a system parameter, we developed a novel technique to pattern concave curvatures into collagen gels. When FTE cells were seeded to confluency on curved substrates, increases in curvature increased the number of invading FTE cells and the invasion distance. FTE invasion into collagen substrates with higher curvature depended on matrix metalloproteinases (MMPs), but expression of collagen I degrading Mmps was not different on curved and flat regions. A finite-element model predicted that contractility and cell–cell connections were essential for increased invasion on substrates with higher curvature, while cell–substrate interactions had minimal effect. Experiments supported these predictions, with invasion decreased by blebbistatin, ethylene glycol-bis(β-aminoethyl ether)-N,N,N’,N’-tetraacetic acid (EGTA) or N-cadherin-blocking antibody, but with no effect from a focal adhesion kinase inhibitor. Finally, experimental evidence supports that cell invasion on curved substrates occurs in two phases—a cell–cell-dependent initiation phase where individual cells break away from the monolayer and an MMP-dependent phase as cells migrate further into the collagen matrix. 
    more » « less
  • Citation Formats
  • MLA
  • APA
  • Chicago
  • BibTeX
  • Save / Share this Record
  • Send to Email