Browsing by Author "Trivedi, Hari"
Now showing 1 - 10 of 12
Results Per Page
Sort Options
Item AI recognition of patient race in medical imaging: a modelling study(Elsevier, 2022-06) Gichoya, Judy Wawira; Banerjee, Imon; Bhimireddy, Ananth Reddy; Burns, John L.; Celi, Leo Anthony; Chen, Li-Ching; Correa, Ramon; Dullerud, Natalie; Ghassemi, Marzyeh; Huang, Shih-Cheng; Kuo, Po-Chih; Lungren, Matthew P.; Palmer, Lyle J.; Price, Brandon J.; Purkayastha, Saptarshi; Pyrros, Ayis T.; Oakden-Rayner, Lauren; Okechukwu, Chima; Seyyed-Kalantari, Laleh; Trivedi, Hari; Wang, Ryan; Zaiman, Zachary; Zhang, Haoran; BioHealth Informatics, School of Informatics and ComputingBackground Previous studies in medical imaging have shown disparate abilities of artificial intelligence (AI) to detect a person's race, yet there is no known correlation for race on medical imaging that would be obvious to human experts when interpreting the images. We aimed to conduct a comprehensive evaluation of the ability of AI to recognise a patient's racial identity from medical images. Methods Using private (Emory CXR, Emory Chest CT, Emory Cervical Spine, and Emory Mammogram) and public (MIMIC-CXR, CheXpert, National Lung Cancer Screening Trial, RSNA Pulmonary Embolism CT, and Digital Hand Atlas) datasets, we evaluated, first, performance quantification of deep learning models in detecting race from medical images, including the ability of these models to generalise to external environments and across multiple imaging modalities. Second, we assessed possible confounding of anatomic and phenotypic population features by assessing the ability of these hypothesised confounders to detect race in isolation using regression models, and by re-evaluating the deep learning models by testing them on datasets stratified by these hypothesised confounding variables. Last, by exploring the effect of image corruptions on model performance, we investigated the underlying mechanism by which AI models can recognise race. Findings In our study, we show that standard AI deep learning models can be trained to predict race from medical images with high performance across multiple imaging modalities, which was sustained under external validation conditions (x-ray imaging [area under the receiver operating characteristics curve (AUC) range 0·91-0·99], CT chest imaging [0·87-0·96], and mammography [0·81]). We also showed that this detection is not due to proxies or imaging-related surrogate covariates for race (eg, performance of possible confounders: body-mass index [AUC 0·55], disease distribution [0·61], and breast density [0·61]). Finally, we provide evidence to show that the ability of AI deep learning models persisted over all anatomical regions and frequency spectrums of the images, suggesting the efforts to control this behaviour when it is undesirable will be challenging and demand further study. Interpretation The results from our study emphasise that the ability of AI deep learning models to predict self-reported race is itself not the issue of importance. However, our finding that AI can accurately predict self-reported race, even from corrupted, cropped, and noised medical images, often when clinical experts cannot, creates an enormous risk for all model deployments in medical imaging. Funding National Institute of Biomedical Imaging and Bioengineering, MIDRC grant of National Institutes of Health, US National Science Foundation, National Library of Medicine of the National Institutes of Health, and Taiwan Ministry of Science and Technology.Item Current Clinical Applications of Artificial Intelligence in Radiology and Their Best Supporting Evidence(Elsevier, 2020-11) Tariq, Amara; Purkayastha, Saptarshi; Padmanaban, Geetha Priya; Krupinski, Elizabeth; Trivedi, Hari; Banerjee, Imon; Gichoya, Judy W.; BioHealth Informatics, School of Informatics and ComputingPurpose Despite tremendous gains from deep learning and the promise of artificial intelligence (AI) in medicine to improve diagnosis and save costs, there exists a large translational gap to implement and use AI products in real-world clinical situations. Adoption of standards such as Transparent Reporting of a Multivariable Prediction Model for Individual Prognosis or Diagnosis, Consolidated Standards of Reporting Trials, and the Checklist for Artificial Intelligence in Medical Imaging is increasing to improve the peer-review process and reporting of AI tools. However, no such standards exist for product-level review. Methods A review of clinical trials showed a paucity of evidence for radiology AI products; thus, the authors developed a 10-question assessment tool for reviewing AI products with an emphasis on their validation and result dissemination. The assessment tool was applied to commercial and open-source algorithms used for diagnosis to extract evidence on the clinical utility of the tools. Results There is limited technical information on methodologies for FDA-approved algorithms compared with open-source products, likely because of intellectual property concerns. Furthermore, FDA-approved products use much smaller data sets compared with open-source AI tools, because the terms of use of public data sets are limited to academic and noncommercial entities, which precludes their use in commercial products. Conclusions Overall, this study reveals a broad spectrum of maturity and clinical use of AI products, but a large gap exists in exploring actual performance of AI tools in clinical practice.Item A DICOM Framework for Machine Learning and Processing Pipelines Against Real-time Radiology Images(SpringerLink, 2021-08-17) Kathiravelu, Pradeeban; Sharma, Puneet; Sharma, Ashish; Banerjee, Imon; Trivedi, Hari; Purkayastha, Saptarshi; Sinha, Priyanshu; Cadrin‑Chenevert, Alexandre; Safdar, Nabile; Wawira Gichoya, Judy; BioHealth Informatics, School of Informatics and ComputingReal-time execution of machine learning (ML) pipelines on radiology images is difficult due to limited computing resources in clinical environments, whereas running them in research clusters requires efficient data transfer capabilities. We developed Niffler, an open-source Digital Imaging and Communications in Medicine (DICOM) framework that enables ML and processing pipelines in research clusters by efficiently retrieving images from the hospitals’ PACS and extracting the metadata from the images. We deployed Niffler at our institution (Emory Healthcare, the largest healthcare network in the state of Georgia) and retrieved data from 715 scanners spanning 12 sites, up to 350 GB/day continuously in real-time as a DICOM data stream over the past 2 years. We also used Niffler to retrieve images bulk on-demand based on user-provided filters to facilitate several research projects. This paper presents the architecture and three such use cases of Niffler. First, we executed an IVC filter detection and segmentation pipeline on abdominal radiographs in real-time, which was able to classify 989 test images with an accuracy of 96.0%. Second, we applied the Niffler Metadata Extractor to understand the operational efficiency of individual MRI systems based on calculated metrics. We benchmarked the accuracy of the calculated exam time windows by comparing Niffler against the Clinical Data Warehouse (CDW). Niffler accurately identified the scanners’ examination timeframes and idling times, whereas CDW falsely depicted several exam overlaps due to human errors. Third, with metadata extracted from the images by Niffler, we identified scanners with misconfigured time and reconfigured five scanners. Our evaluations highlight how Niffler enables real-time ML and processing pipelines in a research cluster.Item MedShift: Automated Identification of Shift Data for Medical Image Dataset Curation(IEEE, 2023) Guo, Xiaoyuan; Wawira Gichoya, Judy; Trivedi, Hari; Purkayastha, Saptarshi; Banerjee, Imon; Biomedical Engineering and Informatics, Luddy School of Informatics, Computing, and EngineeringAutomated curation of noisy external data in the medical domain has long been demanding as AI technologies should be validated on various sources with clean annotated data. To curate a high-quality dataset, identifying variance between the internal and external sources is a fundamental step as the data distributions from different sources can vary significantly and subsequently affect the performance of the AI models. Primary challenges for detecting data shifts are – (1) access to private data across healthcare institutions for manual detection, and (2) the lack of automated approaches to learn efficient shift-data representation without training samples. To overcome the problems, we propose an automated pipeline called MedShift to detect the top-level shift samples and evaluating the significance of shift data without sharing data between the internal and external organizations. MedShift employs unsupervised anomaly detectors to learn the internal distribution and identify samples showing significant shiftness for external datasets, and compared their performance. To quantify the effects of detected shift data, we train a multi-class classifier that learns internal domain knowledge and evaluating the classification performance for each class in external domains after dropping the shift data. We also propose a data quality metric to quantify the dissimilarity between the internal and external datasets. We verify the efficacy of MedShift with musculoskeletal radiographs (MURA) and chest X-rays datasets from more than one external source. Experiments show our proposed shift data detection pipeline can be beneficial for medical centers to curate high-quality datasets more efficiently. The code can be found at https://github.com/XiaoyuanGuo/MedShift. An interface introduction video to visualize our results is available at https://youtu.be/V3BF0P1sxQE.Item MedShift: identifying shift data for medical dataset curation(2021) Guo, Xiaoyuan; Gichoya, Judy Wawira; Trivedi, Hari; Purkayastha, Saptarshi; Banerjee, Imon; BioHealth Informatics, School of Informatics and ComputingTo curate a high-quality dataset, identifying data variance between the internal and external sources is a fundamental and crucial step. However, methods to detect shift or variance in data have not been significantly researched. Challenges to this are the lack of effective approaches to learn dense representation of a dataset and difficulties of sharing private data across medical institutions. To overcome the problems, we propose a unified pipeline called MedShift to detect the top-level shift samples and thus facilitate the medical curation. Given an internal dataset A as the base source, we first train anomaly detectors for each class of dataset A to learn internal distributions in an unsupervised way. Second, without exchanging data across sources, we run the trained anomaly detectors on an external dataset B for each class. The data samples with high anomaly scores are identified as shift data. To quantify the shiftness of the external dataset, we cluster B's data into groups class-wise based on the obtained scores. We then train a multi-class classifier on A and measure the shiftness with the classifier's performance variance on B by gradually dropping the group with the largest anomaly score for each class. Additionally, we adapt a dataset quality metric to help inspect the distribution differences for multiple medical sources. We verify the efficacy of MedShift with musculoskeletal radiographs (MURA) and chest X-rays datasets from more than one external source. Experiments show our proposed shift data detection pipeline can be beneficial for medical centers to curate high-quality datasets more efficiently. An interface introduction video to visualize our results is available at https://youtu.be/V3BF0P1sxQE.Item Multi-Label Medical Image Retrieval Via Learning Multi-Class Similarity(SSRN, 2022) Guo, Xiaoyuan; Duan, Jiali; Gichoya, Judy Wawira; Trivedi, Hari; Purkayastha, Saptarshi; Sharma, Ashish; Banerjee, Imon; BioHealth Informatics, School of Informatics and ComputingIntroduction: Multi-label image retrieval is a challenging problem in the medical area. First, compared to natural images, labels in the medical domain exhibit higher class-imbalance and much nuanced variations. Second, pair-based sampling for positives and negatives during similarity optimization are ambiguous in the multi-label setting, as samples with the same set of labels are limited. Methods: To address the aforementioned challenges, we propose a proxy-based multi-class similarity (PMS) framework, which compares and contrasts samples by comparing their similarities with the discovered proxies. In this way, samples of different sets of label attributes can be utilized and compared indirectly, without the need for complicated sampling. PMS learns a class-wise feature decomposition and maintains a memory bank for positive features from each class. The memory bank keeps track of the latest features, used to compute the class proxies. We compare samples based on their similarity distributions against the proxies, which provide a more stable mean against noise. Results: We benchmark over 10 popular metric learning baselines on two public chest X-ray datasets and experiments show consistent stability of our approach under both exact and non-exact match settings. Conclusions: We proposed a methodology for multi-label medical image retrieval and design a proxy-based multi-class similarity metric, which compares and contrasts samples based on their similarity distributions with respect to the class proxies. With no perquisites, the metrics can be applied to various multi-label medical image applications. The implementation code repository will be publicly available after acceptance.Item Multireader evaluation of radiologist performance for COVID-19 detection on emergency department chest radiographs(Elsevier, 2022-02) Gichoya, Judy W.; Sinha, Priyanshu; Davis, Melissa; Dunkle, Jeffrey W.; Hamlin, Scott A.; Herr, Keith D.; Hoff, Carrie N.; Letter, Haley P.; McAdams, Christopher R.; Puthoff, Gregory D.; Smith, Kevin L.; Steenburg, Scott D.; Banerjee, Imon; Trivedi, Hari; Radiology and Imaging Sciences, School of MedicineBACKGROUND: Chest radiographs (CXR) are frequently used as a screening tool for patients with suspected COVID-19 infection pending reverse transcriptase polymerase chain reaction (RT-PCR) results, despite recommendations against this. We evaluated radiologist performance for COVID-19 diagnosis on CXR at the time of patient presentation in the Emergency Department (ED). MATERIALS AND METHODS: We extracted RT-PCR results, clinical history, and CXRs of all patients from a single institution between March and June 2020. 984 RT-PCR positive and 1043 RT-PCR negative radiographs were reviewed by 10 emergency radiologists from 4 academic centers. 100 cases were read by all radiologists and 1927 cases by 2 radiologists. Each radiologist chose the single best label per case: Normal, COVID-19, Other - Infectious, Other - Noninfectious, Non-diagnostic, and Endotracheal Tube. Cases labeled with endotracheal tube (246) or non-diagnostic (54) were excluded. Remaining cases were analyzed for label distribution, clinical history, and inter-reader agreement. RESULTS: 1727 radiographs (732 RT-PCR positive, 995 RT-PCR negative) were included from 1594 patients (51.2% male, 48.8% female, age 59 ± 19 years). For 89 cases read by all readers, there was poor agreement for RT-PCR positive (Fleiss Score 0.36) and negative (Fleiss Score 0.46) exams. Agreement between two readers on 1638 cases was 54.2% (373/688) for RT-PCR positive cases and 71.4% (679/950) for negative cases. Agreement was highest for RT-PCR negative cases labeled as Normal (50.4%, n = 479). Reader performance did not improve with clinical history or time between CXR and RT-PCR result. CONCLUSION: At the time of presentation to the emergency department, emergency radiologist performance is non-specific for diagnosing COVID-19.Item OSCARS: An Outlier-Sensitive Content-Based Radiography Retrieval System(arXiv, 2022) Guo, Xiaoyuan; Duan, Jiali; Purkayastha, Saptarshi; Trivedi, Hari; Gichoya, Judy Wawira; Banerjee, Imon; BioHealth Informatics, School of Informatics and ComputingImproving the retrieval relevance on noisy datasets is an emerging need for the curation of a large-scale clean dataset in the medical domain. While existing methods can be applied for class-wise retrieval (aka. inter-class), they cannot distinguish the granularity of likeness within the same class (aka. intra-class). The problem is exacerbated on medical external datasets, where noisy samples of the same class are treated equally during training. Our goal is to identify both intra/inter-class similarities for fine-grained retrieval. To achieve this, we propose an Outlier-Sensitive Content-based rAdiologhy Retrieval System (OSCARS), consisting of two steps. First, we train an outlier detector on a clean internal dataset in an unsupervised manner. Then we use the trained detector to generate the anomaly scores on the external dataset, whose distribution will be used to bin intra-class variations. Second, we propose a quadruplet (a, p, nintra, ninter) sampling strategy, where intra-class negatives nintra are sampled from bins of the same class other than the bin anchor a belongs to, while niner are randomly sampled from inter-classes. We suggest a weighted metric learning objective to balance the intra and inter-class feature learning. We experimented on two representative public radiography datasets. Experiments show the effectiveness of our approach.Item Patient-specific COVID-19 resource utilization prediction using fusion AI model(Springer Nature, 2021-06-03) Tariq, Amara; Celi, Leo Anthony; Newsome, Janice M.; Purkayastha, Saptarshi; Bhatia, Neal Kumar; Trivedi, Hari; Wawira Gichoya, Judy; Banerjee, Imon; BioHealth Informatics, School of Informatics and ComputingThe strain on healthcare resources brought forth by the recent COVID-19 pandemic has highlighted the need for efficient resource planning and allocation through the prediction of future consumption. Machine learning can predict resource utilization such as the need for hospitalization based on past medical data stored in electronic medical records (EMR). We conducted this study on 3194 patients (46% male with mean age 56.7 (±16.8), 56% African American, 7% Hispanic) flagged as COVID-19 positive cases in 12 centers under Emory Healthcare network from February 2020 to September 2020, to assess whether a COVID-19 positive patient’s need for hospitalization can be predicted at the time of RT-PCR test using the EMR data prior to the test. Five main modalities of EMR, i.e., demographics, medication, past medical procedures, comorbidities, and laboratory results, were used as features for predictive modeling, both individually and fused together using late, middle, and early fusion. Models were evaluated in terms of precision, recall, F1-score (within 95% confidence interval). The early fusion model is the most effective predictor with 84% overall F1-score [CI 82.1–86.1]. The predictive performance of the model drops by 6 % when using recent clinical data while omitting the long-term medical history. Feature importance analysis indicates that history of cardiovascular disease, emergency room visits in the past year prior to testing, and demographic factors are predictive of the disease trajectory. We conclude that fusion modeling using medical history and current treatment data can forecast the need for hospitalization for patients infected with COVID-19 at the time of the RT-PCR test.Item Reading Race: AI Recognises Patient's Racial Identity In Medical Images(arXiv, 2021) Banerjee, Imon; Bhimireddy, Ananth Reddy; Burns, John L.; Celi, Leo Anthony; Chen, Li-Ching; Correa, Ramon; Dullerud, Natalie; Ghassemi, Marzyeh; Huang, Shih-Cheng; Kuo, Po-Chih; Lungren, Matthew P.; Palmer, Lyle; Price, Brandon J.; Purkayastha, Saptarshi; Pyrros, Ayis; Oakden-Rayner, Luke; Okechukwu, Chima; Seyyed-Kalantari, Laleh; Trivedi, Hari; Wang, Ryan; Zaiman, Zachary; Zhang, Haoran; Gichoya, Judy W.; BioHealth Informatics, School of Informatics and ComputingBackground: In medical imaging, prior studies have demonstrated disparate AI performance by race, yet there is no known correlation for race on medical imaging that would be obvious to the human expert interpreting the images. Methods: Using private and public datasets we evaluate: A) performance quantification of deep learning models to detect race from medical images, including the ability of these models to generalize to external environments and across multiple imaging modalities, B) assessment of possible confounding anatomic and phenotype population features, such as disease distribution and body habitus as predictors of race, and C) investigation into the underlying mechanism by which AI models can recognize race. Findings: Standard deep learning models can be trained to predict race from medical images with high performance across multiple imaging modalities. Our findings hold under external validation conditions, as well as when models are optimized to perform clinically motivated tasks. We demonstrate this detection is not due to trivial proxies or imaging-related surrogate covariates for race, such as underlying disease distribution. Finally, we show that performance persists over all anatomical regions and frequency spectrum of the images suggesting that mitigation efforts will be challenging and demand further study. Interpretation: We emphasize that model ability to predict self-reported race is itself not the issue of importance. However, our findings that AI can trivially predict self-reported race -- even from corrupted, cropped, and noised medical images -- in a setting where clinical experts cannot, creates an enormous risk for all model deployments in medical imaging: if an AI model secretly used its knowledge of self-reported race to misclassify all Black patients, radiologists would not be able to tell using the same data the model has access to.