skip to main content

Search for: All records

Award ID contains: 1908299

Note: When clicking on a Digital Object Identifier (DOI) number, you will be taken to an external site maintained by the publisher. Some full text articles may not yet be available without a charge during the embargo (administrative interval).
What is a DOI Number?

Some links on this page may take you to non-federal websites. Their policies may differ from this site.

  1. Wang, L. ; Dou, Q. ; Fletcher, P.T. ; Speidel, S. ; Li, S. (Ed.)
    Model calibration measures the agreement between the predicted probability estimates and the true correctness likelihood. Proper model calibration is vital for high-risk applications. Unfortunately, modern deep neural networks are poorly calibrated, compromising trustworthiness and reliability. Medical image segmentation particularly suffers from this due to the natural uncertainty of tissue boundaries. This is exasperated by their loss functions, which favor overconfidence in the majority classes. We address these challenges with DOMINO, a domain-aware model calibration method that leverages the semantic confusability and hierarchical similarity between class labels. Our experiments demonstrate that our DOMINO-calibrated deep neural networks outperform non-calibrated models and state-of-the-art morphometric methods in head image segmentation. Our results show that our method can consistently achieve better calibration, higher accuracy, and faster inference times than these methods, especially on rarer classes. This performance is attributed to our domain-aware regularization to inform semantic model calibration. These findings show the importance of semantic ties between class labels in building confidence in deep learning models. The framework has the potential to improve the trustworthiness and reliability of generic medical image segmentation models. The code for this article is available at: https://github.com/lab-smile/DOMINO.
    Free, publicly-accessible full text available September 16, 2023
  2. Abstract Background Diabetic retinopathy (DR) is a leading cause of blindness in American adults. If detected, DR can be treated to prevent further damage causing blindness. There is an increasing interest in developing artificial intelligence (AI) technologies to help detect DR using electronic health records. The lesion-related information documented in fundus image reports is a valuable resource that could help diagnoses of DR in clinical decision support systems. However, most studies for AI-based DR diagnoses are mainly based on medical images; there is limited studies to explore the lesion-related information captured in the free text image reports. Methods In this study, we examined two state-of-the-art transformer-based natural language processing (NLP) models, including BERT and RoBERTa, compared them with a recurrent neural network implemented using Long short-term memory (LSTM) to extract DR-related concepts from clinical narratives. We identified four different categories of DR-related clinical concepts including lesions, eye parts, laterality, and severity, developed annotation guidelines, annotated a DR-corpus of 536 image reports, and developed transformer-based NLP models for clinical concept extraction and relation extraction. We also examined the relation extraction under two settings including ‘gold-standard’ setting—where gold-standard concepts were used–and end-to-end setting. Results For concept extraction, the BERT model pretrained withmore »the MIMIC III dataset achieve the best performance (0.9503 and 0.9645 for strict/lenient evaluation). For relation extraction, BERT model pretrained using general English text achieved the best strict/lenient F1-score of 0.9316. The end-to-end system, BERT_general_e2e, achieved the best strict/lenient F1-score of 0.8578 and 0.8881, respectively. Another end-to-end system based on the RoBERTa architecture, RoBERTa_general_e2e, also achieved the same performance as BERT_general_e2e in strict scores. Conclusions This study demonstrated the efficiency of transformer-based NLP models for clinical concept extraction and relation extraction. Our results show that it’s necessary to pretrain transformer models using clinical text to optimize the performance for clinical concept extraction. Whereas, for relation extraction, transformers pretrained using general English text perform better.« less
    Free, publicly-accessible full text available September 1, 2023
  3. A body of studies has proposed to obtain high-quality images from low-dose and noisy Computed Tomography (CT) scans for radiation reduction. However, these studies are designed for population-level data without considering the variation in CT devices and individuals, limiting the current approaches' performance, especially for ultra-low-dose CT imaging. Here, we proposed PIMA-CT, a physical anthropomorphic phantom model integrating an unsupervised learning framework, using a novel deep learning technique called Cyclic Simulation and Denoising (CSD), to address these limitations. We first acquired paired low-dose and standard-dose CT scans of the phantom and then developed two generative neural networks: noise simulator and denoiser. The simulator extracts real low-dose noise and tissue features from two separate image spaces (e.g., low-dose phantom model scans and standard-dose patient scans) into a unified feature space. Meanwhile, the denoiser provides feedback to the simulator on the quality of the generated noise. In this way, the simulator and denoiser cyclically interact to optimize network learning and ease the denoiser to simultaneously remove noise and restore tissue features. We thoroughly evaluate our method for removing both real low-dose noise and Gaussian simulated low-dose noise. The results show that CSD outperforms one of the state-of-the-art denoising algorithms without using anymore »labeled data (actual patients' low-dose CT scans) nor simulated low-dose CT scans. This study may shed light on incorporating physical models in medical imaging, especially for ultra-low level dose CT scans restoration.« less
    Free, publicly-accessible full text available May 25, 2023
  4. Free, publicly-accessible full text available January 1, 2023
  5. Background and Objectives: Prediction of decline to dementia using objective biomarkers in high-risk patients with amnestic mild cognitive impairment (aMCI) has immense utility. Our objective was to use multimodal MRI to (1) determine whether accurate and precise prediction of dementia conversion could be achieved using baseline data alone, and (2) generate a map of the brain regions implicated in longitudinal decline to dementia. Methods: Participants meeting criteria for aMCI at baseline ( N = 55) were classified at follow-up as remaining stable/improved in their diagnosis ( N = 41) or declined to dementia ( N = 14). Baseline T1 structural MRI and resting-state fMRI (rsfMRI) were combined and a semi-supervised support vector machine (SVM) which separated stable participants from those who decline at follow-up with maximal margin. Cross-validated model performance metrics and MRI feature weights were calculated to include the strength of each brain voxel in its ability to distinguish the two groups. Results: Total model accuracy for predicting diagnostic change at follow-up was 92.7% using baseline T1 imaging alone, 83.5% using rsfMRI alone, and 94.5% when combining T1 and rsfMRI modalities. Feature weights that survived the p < 0.01 threshold for separation of the two groups revealed the strongestmore »margin in the combined structural and functional regions underlying the medial temporal lobes in the limbic system. Discussion: An MRI-driven SVM model demonstrates accurate and precise prediction of later dementia conversion in aMCI patients. The multi-modal regions driving this prediction were the strongest in the medial temporal regions of the limbic system, consistent with literature on the progression of Alzheimer’s disease.« less
    Free, publicly-accessible full text available December 7, 2022
  6. Abstract Alzheimer's disease is the leading cause of dementia. The long progression period in Alzheimer's disease provides a possibility for patients to get early treatment by having routine screenings. However, current clinical diagnostic imaging tools do not meet the specific requirements for screening procedures due to high cost and limited availability. In this work, we took the initiative to evaluate the retina, especially the retinal vasculature, as an alternative for conducting screenings for dementia patients caused by Alzheimer's disease. Highly modular machine learning techniques were employed throughout the whole pipeline. Utilizing data from the UK Biobank, the pipeline achieved an average classification accuracy of 82.44%. Besides the high classification accuracy, we also added a saliency analysis to strengthen this pipeline's interpretability. The saliency analysis indicated that within retinal images, small vessels carry more information for diagnosing Alzheimer's diseases, which aligns with related studies.
  7. Introduction: Computed tomography perfusion (CTP) imaging requires injection of an intravenous contrast agent and increased exposure to ionizing radiation. This process can be lengthy, costly, and potentially dangerous to patients, especially in emergency settings. We propose MAGIC, a multitask, generative adversarial network-based deep learning model to synthesize an entire CTP series from only a non-contrasted CT (NCCT) input. Materials and Methods: NCCT and CTP series were retrospectively retrieved from 493 patients at UF Health with IRB approval. The data were deidentified and all images were resized to 256x256 pixels. The collected perfusion data were analyzed using the RapidAI CT Perfusion analysis software (iSchemaView, Inc. CA) to generate each CTP map. For each subject, 10 CTP slices were selected. Each slice was paired with one NCCT slice at the same location and two NCCT slices at a predefined vertical offset, resulting in 4.3K CTP images and 12.9K NCCT images used for training. The incorporation of a spatial offset into the NCCT input allows MAGIC to more accurately synthesize cerebral perfusive structures, increasing the quality of the generated images. The studies included a variety of indications, including healthy tissue, mild infarction, and severe infarction. The proposed MAGIC model incorporates a novel multitaskmore »architecture, allowing for the simultaneous synthesis of four CTP modalities: mean transit time (MTT), cerebral blood flow (CBF), cerebral blood volume (CBV), and time to peak (TTP). We propose a novel Physicians-in-the-loop module in the model's architecture, acting as a tunable layer that allows physicians to manually adjust the amount of anatomic detail present in the synthesized CTP series. Additionally, we propose two novel loss terms: multi-modal connectivity loss and extrema loss. The multi-modal connectivity loss leverages the multi-task nature to assert that the mathematical relationship between MTT, CBF, and CBV is satisfied. The extrema loss aids in learning regions of elevated and decreased activity in each modality, allowing for MAGIC to accurately learn the characteristics of diagnostic regions of interest. Corresponding NCCT and CTP slices were paired along the vertical axis. The model was trained for 100 epochs on a NVIDIA TITAN X GPU. Results and Discussion: The MAGIC model’s performance was evaluated on a sample of 40 patients from the UF Health dataset. Across all CTP modalities, MAGIC was able to accurately produce images with high structural agreement between the entire synthesized and clinical perfusion images (SSIMmean=0.801 , UQImean=0.926). MAGIC was able to synthesize CTP images to accurately characterize cerebral circulatory structures and identify regions of infarct tissue, as shown in Figure 1. A blind binary evaluation was conducted to assess the presence of cerebral infarction in both the synthesized and clinical perfusion images, resulting in the synthesized images correctly predicting the presence of cerebral infarction with 87.5% accuracy. Conclusions: We proposed a MAGIC model whose novel deep learning structures and loss terms enable high-quality synthesis of CTP maps and characterization of circulatory structures solely from NCCT images, potentially eliminating the requirement for the injection of an intravenous contrast agent and elevated radiation exposure during perfusion imaging. This makes MAGIC a beneficial tool in a clinical scenario increasing the overall safety, accessibility, and efficiency of cerebral perfusion and facilitating better patient outcomes. Acknowledgements: This work was partially supported by the National Science Foundation, IIS-1908299 III: Small: Modeling Multi-Level Connectivity of Brain Dynamics + REU Supplement, to the University of Florida.« less
  8. Introduction: Back pain is one of the most common causes of pain in the United States. Spinal cord stimulation (SCS) is an intervention for patients with chronic back pain (CBP). However, SCS decreases pain in only 58% of patients and relies on self-reported pain scores as outcome measures. An SCS trial is temporarily implanted for seven days and helps to determine if a permanent SCS is needed. Patients that have a >50% reduction in pain from the trial stimulator makes them eligible for permanent implantation. However, self-reported measures reveal little on how mechanisms in the brain are altered. Other measurements of pain intensity, onset, medication, disabilities, depression, and anxiety have been used with machine learning to predict outcomes with accuracies <70%. We aim to predict long-term SCS responders at 6-months using baseline resting EEG and machine learning. Materials and Methods: We obtained 10-minutes of resting electroencephalography (EEG) and pain questionnaires from nine participants with CBP at two time points: 1) pre-trial baseline. 2) Six months after SCS permanent implant surgery. Subjects were designated as high or moderate responders based on the amount of pain relief provided by the long-term (post six months) SCS, and pain scored on a scale ofmore »0-10 with 0 being no pain and 10 intolerable. We used the resting EEG from baseline to predict long-term treatment outcome. Resting EEG data was fed through a pipeline for classification and to map dipole sources. EEG signals were preprocessed using the EEGLAB toolbox. Independent component analysis and dipole fitting were used to linearly unmix the signal and to map dipole sources from the brain. Spectral analysis was performed to obtain the frequency distribution of the signal. Each power band, delta (1-4 Hz), theta (4-8 Hz), alpha (8-13 Hz), beta (13-30 Hz), and gamma (30-100 Hz), as well as the entire spectrum (1-100 Hz), were used for classification. Furthermore, dipole sources were ranked based on classification feature weights to determine the significance of specific regions in the brain. We used support vector machines to predict pain outcomes. Results and Discussion: We found higher frequency powerbands provide overall classification performance of 88.89%. Differences in power are seen between moderate and high responders in both the frontal and parietal regions for theta, alpha, beta, and the entire spectrum (Fig.1). This can potentially be used to predict patient response to SCS. Conclusions: We found evidence of decreased power in theta, alpha, beta, and entire spectrum in the anterior regions of the parietal cortex and posterior regions of the frontal cortex between moderate and high responders, which can be used for predicting treatment outcomes in long-term pain relief from SCS. Long-term treatment outcome prediction using baseline EEG data has the potential to contribute to decision making in terms of permanent surgery, forgo trial periods, and improve clinical efficiency by beginning to understand the mechanism of action of SCS in the human brain.« less
  9. Introduction: Alzheimer’s disease (AD) causes progressive irreversible cognitive decline and is the leading cause of dementia. Therefore, a timely diagnosis is imperative to maximize neurological preservation. However, current treatments are either too costly or limited in availability. In this project, we explored using retinal vasculature as a potential biomarker for early AD diagnosis. This project focuses on stage 3 of a three-stage modular machine learning pipeline which consisted of image quality selection, vessel map generation, and classification [1]. The previous model only used support vector machine (SVM) to classify AD labels which limited its accuracy to 82%. In this project, random forest and gradient boosting were added and, along with SVM, combined into an ensemble classifier, raising the classification accuracy to 89%. Materials and Methods: Subjects classified as AD were those who were diagnosed with dementia in “Dementia Outcome: Alzheimer’s disease” from the UK Biobank Electronic Health Records. Five control groups were chosen with a 5:1 ratio of control to AD patients where the control patients had the same age, gender, and eye side image as the AD patient. In total, 122 vessel images from each group (AD and control) were used. The vessel maps were then segmented from fundusmore »images through U-net. A t-test feature selection was first done on the training folds and the selected features was fed into the classifiers with a p-value threshold of 0.01. Next, 20 repetitions of 5-fold cross validation were performed where the hyperparameters were solely tuned on the training data. An ensemble classifier consisting of SVM, gradient boosting tree, and random forests was built and the final prediction was made through majority voting and evaluated on the test set. Results and Discussion: Through ensemble classification, accuracy increased by 4-12% relative to the individual classifiers, precision by 9-15%, sensitivity by 2-9%, specificity by at least 9-16%, and F1 score by 712%. Conclusions: Overall, a relatively high classification accuracy was achieved using machine learning ensemble classification with SVM, random forest, and gradient boosting. Although the results are very promising, a limitation of this study is that the requirement of needing images of sufficient quality decreased the amount of control parameters that can be implemented. However, through retinal vasculature analysis, this project shows machine learning’s high potential to be an efficient, more cost-effective alternative to diagnosing Alzheimer’s disease. Clinical Application: Using machine learning for AD diagnosis through retinal images will make screening available for a broader population by being more accessible and cost-efficient. Mobile device based screening can also be enabled at primary screening in resource-deprived regions. It can provide a pathway for future understanding of the association between biomarkers in the eye and brain.« less