-
Robust Automatic Whole Brain Extraction on Magnetic Resonance Imaging of Brain Tumor Patients using Dense-Vnet
Authors:
Sara Ranjbar,
Kyle W. Singleton,
Lee Curtin,
Cassandra R. Rickertsen,
Lisa E. Paulson,
Leland S. Hu,
J. Ross Mitchell,
Kristin R. Swanson
Abstract:
Whole brain extraction, also known as skull strip**, is a process in neuroimaging in which non-brain tissue such as skull, eyeballs, skin, etc. are removed from neuroimages. Skull stri** is a preliminary step in presurgical planning, cortical reconstruction, and automatic tumor segmentation. Despite a plethora of skull strip** approaches in the literature, few are sufficiently accurate for p…
▽ More
Whole brain extraction, also known as skull strip**, is a process in neuroimaging in which non-brain tissue such as skull, eyeballs, skin, etc. are removed from neuroimages. Skull stri** is a preliminary step in presurgical planning, cortical reconstruction, and automatic tumor segmentation. Despite a plethora of skull strip** approaches in the literature, few are sufficiently accurate for processing pathology-presenting MRIs, especially MRIs with brain tumors. In this work we propose a deep learning approach for skull stri** common MRI sequences in oncology such as T1-weighted with gadolinium contrast (T1Gd) and T2-weighted fluid attenuated inversion recovery (FLAIR) in patients with brain tumors. We automatically created gray matter, white matter, and CSF probability masks using SPM12 software and merged the masks into one for a final whole-brain mask for model training. Dice agreement, sensitivity, and specificity of the model (referred herein as DeepBrain) was tested against manual brain masks. To assess data efficiency, we retrained our models using progressively fewer training data examples and calculated average dice scores on the test set for the models trained in each round. Further, we tested our model against MRI of healthy brains from the LBP40A dataset. Overall, DeepBrain yielded an average dice score of 94.5%, sensitivity of 96.4%, and specificity of 98.5% on brain tumor data. For healthy brains, model performance improved to a dice score of 96.2%, sensitivity of 96.6% and specificity of 99.2%. The data efficiency experiment showed that, for this specific task, comparable levels of accuracy could have been achieved with as few as 50 training samples. In conclusion, this study demonstrated that a deep learning model trained on minimally processed automatically-generated labels can generate more accurate brain masks on MRI of brain tumor patients within seconds.
△ Less
Submitted 3 June, 2020;
originally announced June 2020.
-
Sex differences in predicting fluid intelligence of adolescent brain from T1-weighted MRIs
Authors:
Sara Ranjbar,
Kyle W. Singleton,
Lee Curtin,
Susan Christine Massey,
Andrea Hawkins-Daarud,
Pamela R. Jackson,
Kristin R. Swanson
Abstract:
Fluid intelligence (Gf) has been defined as the ability to reason and solve previously unseen problems. Links to Gf have been found in magnetic resonance imaging (MRI) sequences such as functional MRI and diffusion tensor imaging. As part of the Adolescent Brain Cognitive Development Neurocognitive Prediction Challenge 2019, we sought to predict Gf in children aged 9-10 from T1-weighted (T1W) MRIs…
▽ More
Fluid intelligence (Gf) has been defined as the ability to reason and solve previously unseen problems. Links to Gf have been found in magnetic resonance imaging (MRI) sequences such as functional MRI and diffusion tensor imaging. As part of the Adolescent Brain Cognitive Development Neurocognitive Prediction Challenge 2019, we sought to predict Gf in children aged 9-10 from T1-weighted (T1W) MRIs. The data included atlas-aligned volumetric T1W images, atlas-defined segmented regions, age, and sex for 3739 subjects used for training and internal validation and 415 subjects used for external validation. We trained sex-specific convolutional neural net (CNN) and random forest models to predict Gf. For the convolutional model, skull-stripped volumetric T1W images aligned to the SRI24 brain atlas were used for training. Volumes of segmented atlas regions along with each subject's age were used to train the random forest regressor models. Performance was measured using the mean squared error (MSE) of the predictions. Random forest models achieved lower MSEs than CNNs. Further, the external validation data had a better MSE for females than males (60.68 vs. 80.74), with a combined MSE of 70.83. Our results suggest that predictive models of Gf from volumetric T1W MRI features alone may perform better when trained separately on male and female data. However, the performance of our models indicates that more information is necessary beyond the available data to make accurate predictions of Gf.
△ Less
Submitted 6 August, 2019;
originally announced August 2019.
-
A Revision of the Bernoulli Equation as a Controller of the Fick's Diffusion Equation in Drug Delivery Modeling
Authors:
Ali Esmaeili,
Saeed Ranjbar
Abstract:
Mathematical equations can be used as effectual tools in drug delivery systems modeling and are also highly helpful to have a theoretical understanding of controlled drug release and diffusion mechanisms. In this study we aim to present a mathematical combination between the Bernoulli equation and the Fick's equation as a diffusion controller in drug delivery systems. For this propose we have revi…
▽ More
Mathematical equations can be used as effectual tools in drug delivery systems modeling and are also highly helpful to have a theoretical understanding of controlled drug release and diffusion mechanisms. In this study we aim to present a mathematical combination between the Bernoulli equation and the Fick's equation as a diffusion controller in drug delivery systems. For this propose we have revised the Bernoulli equation as an additional, controller and complementary method of the Fick's diffusion equation to detect the optimal delivery direction to control the diffusion divergence of the drug carrier in vascular systems during the transportation process in biological tissues. Therefore, by utilizing the Bernoulli equation we could determine the real direction by the route function f.
△ Less
Submitted 30 January, 2019;
originally announced February 2019.
-
Left ventricular myocardium of heart as a 3D scheme
Authors:
Saeed Ranjbar,
Mersedeh Karvandi
Abstract:
In this paper it will be shown that according to the scheme theory in algebraic geometry, human heart as an elastic body can be represented as a 3D scheme, on account of algebraic equations of the myocardial fibers as the local parts of the global scheme. It is possible that the fiber movements to be discussed here are identical with the so-called "myocardial fiber transactions"; however the infor…
▽ More
In this paper it will be shown that according to the scheme theory in algebraic geometry, human heart as an elastic body can be represented as a 3D scheme, on account of algebraic equations of the myocardial fibers as the local parts of the global scheme. It is possible that the fiber movements to be discussed here are identical with the so-called "myocardial fiber transactions"; however the information available to me regarding the latter is lacking in precision, that I can from no judgment in the matter. It is hoped that some enquirer may succeed shortly in this introduced scheme suggested here, which is so important in connection with the theory of schemes.
△ Less
Submitted 18 June, 2015;
originally announced December 2015.
-
Mathematical modeling applied to the left ventricle of heart
Authors:
Saeed Ranjbar,
Mersedeh Karvandi
Abstract:
Background: How can mathematics help us to understand the mechanism of the cardiac motion? The best known approach is to take a mathematical model of the fibered structure, insert it into a more-or-less complex model of cardiac architecture, and then study the resulting fibers of activation that propagate through the myocardium. In our paper, we have attempted to create a novel software capable of…
▽ More
Background: How can mathematics help us to understand the mechanism of the cardiac motion? The best known approach is to take a mathematical model of the fibered structure, insert it into a more-or-less complex model of cardiac architecture, and then study the resulting fibers of activation that propagate through the myocardium. In our paper, we have attempted to create a novel software capable of demonstrate left ventricular (LV) model in normal hearts.
Method: Echocardiography was performed on 70 healthy volunteers. Data evaluated included: velocity (radial, longitudinal, rotational and vector point), displacement (longitudinal and rotational), strain rate (longitudinal and circumferential) and strain (radial, longitudinal and circumferential) of all 16 LV myocardial segments. Using these data, force vectors of myocardial samples were estimated by MATLAB software, interfaced in the echocardiograph system. Dynamic orientation contraction (through the cardiac cycle) of every individual myocardial fiber could be created by adding together the sequential steps of the multiple fragmented sectors of that fiber. This way we attempted to mechanically illustrate the global LV model.
Result: Our study shows that in normal cases myocardial fibers initiate from the posterior basal region of the heart, continues through the LV free wall, reaches the septum, loops around the apex, ascends, and ends at the superior-anterior edge of LV.
Conclusion: We were able to define the whole LV myocardial model mathematically, by MATLAB software in normal subjects. This will enable physicians to diagnose and follow up many cardiac diseases when this software is interfaced within echocardiographic machines.
△ Less
Submitted 24 June, 2014;
originally announced June 2014.
