Search

Your search keyword '"Musini, Kumudha"' showing total 10 results
Start Over Author "Musini, Kumudha"
10 results on '"Musini, Kumudha"'

2. Generation and Analysis of Feature-Dependent Pseudo Noise for Training Deep Neural Networks

Author

Kamabattula, Sree Ram, Musini, Kumudha, Namazi, Babak, Sankaranarayanan, Ganesh, and Devarajan, Venkat

Subjects
Computer Science - Machine Learning
Abstract

Training Deep neural networks (DNNs) on noisy labeled datasets is a challenging problem, because learning on mislabeled examples deteriorates the performance of the network. As the ground truth availability is limited with real-world noisy datasets, previous papers created synthetic noisy datasets by randomly modifying the labels of training examples of clean datasets. However, no final conclusions can be derived by just using this random noise, since it excludes feature-dependent noise. Thus, it is imperative to generate feature-dependent noisy datasets that additionally provide ground truth. Therefore, we propose an intuitive approach to creating feature-dependent noisy datasets by utilizing the training predictions of DNNs on clean datasets that also retain true label information. We refer to these datasets as "Pseudo Noisy datasets". We conduct several experiments to establish that Pseudo noisy datasets resemble feature-dependent noisy datasets across different conditions. We further randomly generate synthetic noisy datasets with the same noise distribution as that of Pseudo noise (referred as "Randomized Noise") to empirically show that i) learning is easier with feature-dependent label noise compared to random noise, ii) irrespective of noise distribution, Pseudo noisy datasets mimic feature-dependent label noise and iii) current training methods are not generalizable to feature-dependent label noise. Therefore, we believe that Pseudo noisy datasets will be quite helpful to study and develop robust training methods.

Published
2021

5. Quantification of Skeletal Muscle Perfusion in Peripheral Artery Disease Using 18 F‐Sodium Fluoride Positron Emission Tomography Imaging

Author

Chou, Ting‐Heng, primary, Nabavinia, Mahboubeh, additional, Tram, Nguyen K., additional, Rimmerman, Eleanor T., additional, Patel, Surina, additional, Musini, Kumudha Narayana, additional, Eisert, Susan Natalie, additional, Wolfe, Tatiana, additional, Wynveen, Molly K., additional, Matsuzaki, Yuichi, additional, Kitsuka, Takahiro, additional, Iwaki, Ryuma, additional, Janse, Sarah A., additional, Bobbey, Adam J., additional, Breuer, Christopher K., additional, Goodchild, Laurie, additional, Malbrue, Raphael, additional, Shinoka, Toshiharu, additional, Atway, Said A., additional, Go, Michael R., additional, and Stacy, Mitchel R., additional

Published
2024
Full Text
View/download PDF

6. Quantification of Skeletal Muscle Perfusion in Peripheral Artery Disease Using 18F-Sodium Fluoride Positron Emission Tomography Imaging.

Author

Ting-Heng Chou, Nabavinia, Mahboubeh, Tram, Nguyen K., Rimmerman, Eleanor T., Patel, Surina, Musini, Kumudha Narayana, Eisert, Susan Natalie, Wolfe, Tatiana, Wynveen, Molly K., Yuichi Matsuzaki, Takahiro Kitsuka, Ryuma Iwaki, Janse, Sarah A., Bobbey, Adam J., Breuer, Christopher K., Goodchild, Laurie, Malbrue, Raphael, Toshiharu Shinoka, Atway, Said A., and Go, Michael R.

Published
2024
Full Text
View/download PDF

8. Vessel-by-vessel analysis of lower extremity 18F-NaF PET/CT imaging quantifies diabetes- and chronic kidney disease-induced active microcalcification in patients with peripheral arterial disease.

Author

Chou, Ting-Heng, Rimmerman, Eleanor T., Patel, Surina, Wynveen, Molly K., Eisert, Susan N., Musini, Kumudha Narayana, Janse, Sarah A., Bobbey, Adam J., Sarac, Timur P., Atway, Said A., Go, Michael R., and Stacy, Mitchel R.

Subjects
POSITRON emission tomography, PERIPHERAL vascular diseases, COMPUTED tomography, ARTERIAL calcification, CHRONIC kidney failure, IMAGE segmentation
Abstract

Background: Positron emission tomography (PET)/computed tomography (CT) imaging with fluorine-18 (18F)-sodium fluoride (NaF) provides assessment of active vascular microcalcification, but its utility for evaluating diabetes mellitus (DM)- and chronic kidney disease (CKD)-induced atherosclerosis in peripheral arterial disease (PAD) has not been comprehensively evaluated. This study sought to use 18F-NaF PET/CT to quantify and compare active microcalcification on an artery-by-artery basis in healthy subjects, PAD patients with or without DM, and PAD patients with or without CKD. Additionally, we evaluated the contributions of DM, CKD, statin use and established CT-detectable calcium to 18F-NaF uptake for each lower extremity artery. Methods: PAD patients (n = 48) and healthy controls (n = 8) underwent lower extremity 18F-NaF PET/CT imaging. Fused PET/CT images guided segmentation of arteries of interest (i.e., femoral-popliteal, anterior tibial, tibioperoneal trunk, posterior tibial, and peroneal) and quantification of 18F-NaF uptake. 18F-NaF uptake was assessed for each artery and compared between subject groups. Additionally, established calcium burden was quantified for each artery using CT calcium mass score. Univariate and multivariate analyses were performed to evaluate DM, CKD, statin use, and CT calcium mass as predictors of 18F-NaF uptake in PAD. Results: PAD patients with DM or CKD demonstrated significantly higher active microcalcification (i.e., 18F-NaF uptake) for all arteries when compared to PAD patients without DM or CKD. Univariate and multivariate analyses revealed that concomitant DM or CKD was associated with increased microcalcification for all arteries of interest and this increased disease risk remained significant after adjusting for patient age, sex, and body mass index. Statin use was only associated with decreased microcalcification for the femoral-popliteal artery in multivariate analyses. Established CT-detectable calcium was not significantly associated with 18F-NaF uptake for 4 out of 5 arteries of interest. Conclusions: 18F-NaF PET/CT imaging quantifies vessel-specific active microcalcification in PAD that is increased in multiple lower extremity arteries by DM and CKD and decreased in the femoral-popliteal artery by statin use. 18F-NaF PET imaging is complementary to and largely independent of established CT-detectable arterial calcification. 18F-NaF PET/CT imaging may provide an approach for non-invasively quantifying vessel-specific responses to emerging anti-atherogenic therapies or CKD treatment in patients with PAD. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

10. Vessel-by-Vessel CT Calcium Scoring of the Foot in Peripheral Artery Disease: Association with Patient-Level Factors

Author

Rimmerman, Eleanor T., Musini, Kumudha Narayana, Chou, Ting-Heng, Wynveen, Molly K., Patel, Surina A., Beall, Maren, Bobbey, Adam J, Atway, Said A, Go, Michael, and Stacy, Mitchel R.

Abstract

Objective: Peripheral artery disease (PAD) is associated with increased risk of non-healing ulcers, amputation, and mortality due to occlusive atherosclerotic plaques. Computed tomography (CT) imaging detects vascular calcification in PAD; however, quantitative vessel-by-vessel analysis of calcium burden in the feet of PAD patients has not been assessed. This study sought to perform quantitative analysis of vessel-specific calcium burden and examine the patient-level determinants of foot calcium burden in PAD patients. Approach: PAD patients (n=41) were prospectively enrolled and underwent CT imaging of the lower extremities. Manual segmentation of the medial plantar, lateral plantar, and dorsalis pedis arteries was performed. CT image Hounsfield units (HUs) were obtained for each artery to quantify vessel-by-vessel calcium mass using a cutoff value of ≥130 HU. Univariate analyses were performed to evaluate patient-level determinants of calcium burden for each foot artery. STROBE guidelines were used for reporting of data. Results: Univariate analyses revealed that body mass index, diabetes mellitus (DM), and chronic kidney disease (CKD) were significant determinants of foot calcium burden in PAD patients. Image analysis demonstrated that PAD patients with DM had significantly higher calcium mass for the medial plantar (p=0.005), lateral plantar (p=0.039), and dorsalis pedis (p=0.001) arteries compared to PAD patients without DM. Innovation: This is the first study to use CT imaging to quantify vessel-specific calcium burden in the feet of patients with PAD and evaluate the patient-level determinants of foot calcium burden in the setting of PAD. Conclusion: CT imaging quantifies vessel-specific calcification in the feet of PAD patients, which is exacerbated with concomitant DM, CKD, and/or obesity.

Published
2023
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results