The NIH/NIGMS
Center for Integrative Biomedical Computing

SCI Publications

2024


J.A. Bergquist, B. Zenger, J. Brundage, R.S. MacLeod, T.J. Bunch, R. Shah, X. Ye, A. Lyons, M. Torre, R. Ranjan, T. Tasdizen, B.A. Steinberg. “Performance of Off-the-Shelf Machine Learning Architectures and Biases in Low Left Ventricular Ejection Fraction Detection,” In Heart Rhythm O2, Vol. 5, No. 9, pp. 644 - 654. 2024.

ABSTRACT

Background

Artificial intelligence–machine learning (AI-ML) has demonstrated the ability to extract clinically useful information from electrocardiograms (ECGs) not available using traditional interpretation methods. There exists an extensive body of AI-ML research in fields outside of cardiology including several open-source AI-ML architectures that can be translated to new problems in an “off-the-shelf” manner.

Objective

We sought to address the limited investigation of which if any of these off-the-shelf architectures could be useful in ECG analysis as well as how and when these AI-ML approaches fail.

Methods

We applied 6 off-the-shelf AI-ML architectures to detect low left ventricular ejection fraction (LVEF) in a cohort of ECGs from 24,868 patients. We assessed LVEF classification and explored patient characteristics associated with inaccurate (false positive or false negative) LVEF prediction.

Results

We found that all of these network architectures produced LVEF detection area under the receiver-operating characteristic curve values above 0.9 (averaged over 5 instances per network), with the ResNet 18 network performing the highest (average area under the receiver-operating characteristic curve of 0.917). We also observed that some patient-specific characteristics such as race, sex, and presence of several comorbidities were associated with lower LVEF prediction performance.

Conclusions

This demonstrates the ability of off-the-shelf AI-ML architectures to detect clinically useful information from ECGs with performance matching contemporary custom-build AI-ML architectures. We also highlighted the presence of possible biases in these AI-ML approaches in the context of patient characteristics. These findings should be considered in the pursuit of efficient and equitable deployment of AI-ML technologies moving forward.


J.A. Bergquist, D. Dade, B. Zenger, R.S. MacLeod, X. Ye, R. Ranjan, T. Tasdizen, B.A. Steinberg. “Machine Learning Prediction of Blood Potassium at Different Time Cutoffs,” In Computing in Cardiology 2024, 2024.

ABSTRACT

Because serum potassium and ECG morphology changes exhibit a well-understood connection, and the timeline of ECG changes can be relatively quick, there is motivation to explore the sensitivity of ML based prediction of serum potassium using 12 lead ECG data with respect to the time between the ECG and potassium readings.

We trained a convolutional neural network to classify abnormal (serum potassium above 5 mEq/L) vs normal (serum potassium between 4 and 5 mEq/L) from the ECG alone. We compared training with ECGs and potassium measurements filtered to be within 1 hour, 30 minutes, and 15 minutes of each other. We explored scenarios that both leveraged all available data at each time cutoff as well as restricted data to match training set sizes across the time cutoffs. For each case, we trained five separate instances of our neural network to account for variability.

The 1 hour cutoff with all data resulted in an average area under the receiver operator curve (AUC) of 0.850 and a weighted accuracy of 76.3%, 15 minutes resulted in 0.814, 72.5%, and 30 minutes. Truncating the training sets to the same size as the 15 minute cutoff results in comparable average accuracy and AUC for all. Our future studies will continue to explore the performance of ML potassium predictions through investigations of failure cases, identification of biases, and explainability analyses.

2013


C. Jones, M. Seyedhosseini, M. Ellisman, T. Tasdizen. “Neuron Segmentation in Electron Microscopy Images Using Partial Differential Equations,” In Proceedings of 2013 IEEE 10th International Symposium on Biomedical Imaging (ISBI), pp. 1457--1460. April, 2013.
DOI: 10.1109/ISBI.2013.6556809

ABSTRACT

×
In connectomics, neuroscientists seek to identify the synaptic connections between neurons. Segmentation of cell membranes using supervised learning algorithms on electron microscopy images of brain tissue is often done to assist in this effort. Here we present a partial differential equation with a novel growth term to improve the results of a supervised learning algorithm. We also introduce a new method for representing the resulting image that allows for a more dynamic thresholding to further improve the result. Using these two processes we are able to close small to medium sized gaps in the cell membrane detection and improve the Rand error by as much as 9\% over the initial supervised segmentation.

2010


J.R. Anderson, B.C. Grimm, S. Mohammed, B.W. Jones, T. Tasdizen, J. Spaltenstein, P. Koshevoy, R.T. Whitaker, R.E. Marc. “The Viking Viewer: Scalable Multiuser Annotation and Summarization of Large Volume Datasets,” In Journal of Microscopy, Vol. 241, No. 1, pp. 13--28. 2010.
DOI: 10.1111/j.1365-2818.2010.03402.x


S. Gerber, T. Tasdizen, P.T. Fletcher, S. Joshi, R.T. Whitaker, the Alzheimers Disease Neuroimaging Initiative (ADNI). “Manifold modeling for brain population analysis,” In Medical Image Analysis, Special Issue on the 12th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) 2009, Vol. 14, No. 5, Note: Awarded MICCAI 2010, Best of the Journal Issue Award, pp. 643--653. 2010.
ISSN: 1361-8415
DOI: 10.1016/j.media.2010.05.008
PubMed ID: 20579930


S.K. Iyer, E. DiBella, T. Tasdizen. “Edge enhanced spatio-temporal constrained reconstruction of undersampled dynamic contrast enhanced radial MRI,” In IEEE International Symposium on Biomedical Imaging (ISBI): From Nano to Macro, pp. 704--707. 2010.
DOI: 10.1109/ISBI.2010.5490077

2007


G. Adluru, S.P. Awate, T. Tasdizen, R.T. Whitaker, E.V.R. DiBella. “Temporally Constrained Reconstruction of Dynamic Cardiac Perfusion MRI,” In Magnetic Resonance in Medicine, Vol. 57, pp. 1027--1036. 2007.

2006


S.P. Awate, T. Tasdizen, R.T. Whitaker. “Unsupervised Texture Segmentation with Nonparametric Neighborhood Statistics,” SCI Institute Technical Report, No. UUSCI-2006-011, University of Utah, 2006.


E. Jurrus, T. Tasdizen, P. Koshevoy, P.T. Fletcher, M. Hardy, C-B. Chien, W. Denk, R.T. Whitaker. “Axon Tracking in Serial Block-Face Scanning Electron Microscopy,” In Workshop on Microscopic Image Analysis with Applications in Biology, MICCAI, October, 2006.

2005


T. Tasdizen, S.P. Awate, R.T. Whitaker, N. Foster. “MRI Tissue Classification with Neighborhood Statistics: A Nonparametric, Entropy-Minimizing Approach,” In Proceedings of The 8th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI), pp. 517--525. 2005.
PubMed ID: 16685999

2004


T. Tasdizen, D.M. Weinstein, J.N. Lee. “Automatic Tissue Classification for the Human Head from Multispectral MRI,” SCI Institute Technical Report, No. UUSCI-2004-001, University of Utah, March, 2004.


T. Tasdizen, R.T. Whitaker. “Higher-order nonlinear priors for surface reconstruction,” In IEEE Trans. Pattern Anal. & Mach. Intel., Vol. 26, No. 7, pp. 878--891. July, 2004.