Search

Your search keyword '"Rao, Chinmay"' showing total 28 results
Start Over Author "Rao, Chinmay"
28 results on '"Rao, Chinmay"'

1. Improving Uncertainty-Error Correspondence in Deep Bayesian Medical Image Segmentation

Author

Mody, Prerak, Chaves-de-Plaza, Nicolas F., Rao, Chinmay, Astrenidou, Eleftheria, de Ridder, Mischa, Hoekstra, Nienke, Hildebrandt, Klaus, and Staring, Marius

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Artificial Intelligence, Computer Science - Human-Computer Interaction, Computer Science - Machine Learning
Abstract

Increased usage of automated tools like deep learning in medical image segmentation has alleviated the bottleneck of manual contouring. This has shifted manual labour to quality assessment (QA) of automated contours which involves detecting errors and correcting them. A potential solution to semi-automated QA is to use deep Bayesian uncertainty to recommend potentially erroneous regions, thus reducing time spent on error detection. Previous work has investigated the correspondence between uncertainty and error, however, no work has been done on improving the "utility" of Bayesian uncertainty maps such that it is only present in inaccurate regions and not in the accurate ones. Our work trains the FlipOut model with the Accuracy-vs-Uncertainty (AvU) loss which promotes uncertainty to be present only in inaccurate regions. We apply this method on datasets of two radiotherapy body sites, c.f. head-and-neck CT and prostate MR scans. Uncertainty heatmaps (i.e. predictive entropy) are evaluated against voxel inaccuracies using Receiver Operating Characteristic (ROC) and Precision-Recall (PR) curves. Numerical results show that when compared to the Bayesian baseline the proposed method successfully suppresses uncertainty for accurate voxels, with similar presence of uncertainty for inaccurate voxels. Code to reproduce experiments is available at https://github.com/prerakmody/bayesuncertainty-error-correspondence, Comment: Accepted for publication at the Journal of Machine Learning for Biomedical Imaging (MELBA) https://melba-journal.org/2024:018

Published
2024
Full Text
View/download PDF

2. Oropharyngeal Tumour Segmentation Using Ensemble 3D PET-CT Fusion Networks for the HECKTOR Challenge

Author

Rao, Chinmay, Pai, Suraj, Hadzic, Ibrahim, Zhovannik, Ivan, Bontempi, Dennis, Dekker, Andre, Teuwen, Jonas, Traverso, Alberto, Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Woeginger, Gerhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Andrearczyk, Vincent, editor, Oreiller, Valentin, editor, and Depeursinge, Adrien, editor

Published
2021
Full Text
View/download PDF

8. Head and neck tumor segmentation in PET/CT: The HECKTOR challenge

Author

Oreiller, Valentin, primary, Andrearczyk, Vincent, additional, Jreige, Mario, additional, Boughdad, Sarah, additional, Elhalawani, Hesham, additional, Castelli, Joel, additional, Vallières, Martin, additional, Zhu, Simeng, additional, Xie, Juanying, additional, Peng, Ying, additional, Iantsen, Andrei, additional, Hatt, Mathieu, additional, Yuan, Yading, additional, Ma, Jun, additional, Yang, Xiaoping, additional, Rao, Chinmay, additional, Pai, Suraj, additional, Ghimire, Kanchan, additional, Feng, Xue, additional, Naser, Mohamed A., additional, Fuller, Clifton D., additional, Yousefirizi, Fereshteh, additional, Rahmim, Arman, additional, Chen, Huai, additional, Wang, Lisheng, additional, Prior, John O., additional, and Depeursinge, Adrien, additional

Published
2022
Full Text
View/download PDF

25. Nondestructive forensic pathology of Lead-Acid batteries.

Author

Shi, Ying, Ferone, Chris, Rao, Chinmay, and Rahn, Christopher D.

Abstract

Valve Regulated Lead-Acid (VRLA) batteries can degrade due to a variety of mechanisms, including corrosion, hard sulfation, water loss, shedding, and active mass degradation. VRLA batteries are designed to minimize these effects as much as possible but the operating environment, cell-to-cell and battery-to-battery manufacturing variations, and use can cause different degradation mechanisms to dominate capacity loss and/or impedance rise. With accurate State of Health monitoring, cell usage can be adjusted by the battery management system (BMS) to optimize the performance and life of the energy storage system. The BMS must be able to determine in real-time the predominant degradation mechanism for each cell and adjust use accordingly. In this paper, new and dead VRLA batteries are tested with constant, sinusoidal, and pulse charge/discharge current inputs while measuring the cell voltage and pressure to determine the cause of death of the cells. As expected, the new cells have fairly uniform performance with limited signs of degradation. The cells in the dead battery, however, have widely ranging performance, especially at the end of discharge and charge. Analysis of the charge/discharge data indicate that two cells died of water loss and a third cell died of sulfation. The remaining three cells were fairly healthy but will accompany their dead companions to the recycling center nonetheless. While the full charge/discharge data provided useful forensic pathology data, EIS and pulse charge/discharge data varied with aging mechanism but the correlation was unclear. [ABSTRACT FROM PUBLISHER]

Published
2012
Full Text
View/download PDF

28. Automated reduction of non-neuronal signals from intra-cortical microwire array recordings by use of correlation technique.

Author

Paralikar K, Rao C, and Clement RS

Subjects
Animals, Electrodes, Implanted, Pattern Recognition, Automated methods, Rats, Rats, Sprague-Dawley, Reproducibility of Results, Sensitivity and Specificity, Statistics as Topic, Algorithms, Artifacts, Brain physiology, Electroencephalography instrumentation, Electroencephalography methods, Evoked Potentials physiology, Microelectrodes, Neurons physiology
Abstract

Implanted intra-cortical micro-electrode arrays record multi-unit extracellular spike activity that is used in deciphering the neural basis for adaptation, learning, plasticity and as command signal for brain-machine interfaces (BMI). Detection of spike activity is the first step in successful implementation of all the aforementioned applications. However, with awake and behaving animals, micro-electrode arrays typically also record non-neuronal signals induced by the animal's movement, feeding and grooming actions. The spectral and temporal nature of these artifacts is similar to neural spikes, which complicates accurate detection. The distal source and higher strength of non-neuronal signals result in their near simultaneous registration on most electrodes, while neural spiking event is rarely recorded on more than one electrode of an array. This difference is utilized in identifying non-neuronal content from acquired data by performing a correlation analysis. The efficacy of the method is evaluated by comparing outcomes from algorithms that use absolute threshold and Principal Component Analysis (PCA) as a means of identifying neural spikes with the same methods incorporating correlation analysis.

Published
2008
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results