ABSTRACT
OBJECTIVES: To develop and validate an open-source artificial intelligence (AI) algorithm to accurately detect contrast phases in abdominal CT scans. MATERIALS AND METHODS: Retrospective study aimed to develop an AI algorithm trained on 739 abdominal CT exams from 2016 to 2021, from 200 unique patients, covering 1545 axial series. We performed segmentation of five key anatomic structures-aorta, portal vein, inferior vena cava, renal parenchyma, and renal pelvis-using TotalSegmentator, a deep learning-based tool for multi-organ segmentation, and a rule-based approach to extract the renal pelvis. Radiomics features were extracted from the anatomical structures for use in a gradient-boosting classifier to identify four contrast phases: non-contrast, arterial, venous, and delayed. Internal and external validation was performed using the F1 score and other classification metrics, on the external dataset "VinDr-Multiphase CT". RESULTS: The training dataset consisted of 172 patients (mean age, 70 years ± 8, 22% women), and the internal test set included 28 patients (mean age, 68 years ± 8, 14% women). In internal validation, the classifier achieved an accuracy of 92.3%, with an average F1 score of 90.7%. During external validation, the algorithm maintained an accuracy of 90.1%, with an average F1 score of 82.6%. Shapley feature attribution analysis indicated that renal and vascular radiodensity values were the most important for phase classification. CONCLUSION: An open-source and interpretable AI algorithm accurately detects contrast phases in abdominal CT scans, with high accuracy and F1 scores in internal and external validation, confirming its generalization capability. CLINICAL RELEVANCE STATEMENT: Contrast phase detection in abdominal CT scans is a critical step for downstream AI applications, deploying algorithms in the clinical setting, and for quantifying imaging biomarkers, ultimately allowing for better diagnostics and increased access to diagnostic imaging. KEY POINTS: Digital Imaging and Communications in Medicine labels are inaccurate for determining the abdominal CT scan phase. AI provides great help in accurately discriminating the contrast phase. Accurate contrast phase determination aids downstream AI applications and biomarker quantification.
ABSTRACT
Analyzing vast textual data and summarizing key information from electronic health records imposes a substantial burden on how clinicians allocate their time. Although large language models (LLMs) have shown promise in natural language processing (NLP) tasks, their effectiveness on a diverse range of clinical summarization tasks remains unproven. Here we applied adaptation methods to eight LLMs, spanning four distinct clinical summarization tasks: radiology reports, patient questions, progress notes and doctor-patient dialogue. Quantitative assessments with syntactic, semantic and conceptual NLP metrics reveal trade-offs between models and adaptation methods. A clinical reader study with 10 physicians evaluated summary completeness, correctness and conciseness; in most cases, summaries from our best-adapted LLMs were deemed either equivalent (45%) or superior (36%) compared with summaries from medical experts. The ensuing safety analysis highlights challenges faced by both LLMs and medical experts, as we connect errors to potential medical harm and categorize types of fabricated information. Our research provides evidence of LLMs outperforming medical experts in clinical text summarization across multiple tasks. This suggests that integrating LLMs into clinical workflows could alleviate documentation burden, allowing clinicians to focus more on patient care.
Subject(s)
Documentation , Semantics , Humans , Electronic Health Records , Natural Language Processing , Physician-Patient RelationsABSTRACT
Sifting through vast textual data and summarizing key information from electronic health records (EHR) imposes a substantial burden on how clinicians allocate their time. Although large language models (LLMs) have shown immense promise in natural language processing (NLP) tasks, their efficacy on a diverse range of clinical summarization tasks has not yet been rigorously demonstrated. In this work, we apply domain adaptation methods to eight LLMs, spanning six datasets and four distinct clinical summarization tasks: radiology reports, patient questions, progress notes, and doctor-patient dialogue. Our thorough quantitative assessment reveals trade-offs between models and adaptation methods in addition to instances where recent advances in LLMs may not improve results. Further, in a clinical reader study with ten physicians, we show that summaries from our best-adapted LLMs are preferable to human summaries in terms of completeness and correctness. Our ensuing qualitative analysis highlights challenges faced by both LLMs and human experts. Lastly, we correlate traditional quantitative NLP metrics with reader study scores to enhance our understanding of how these metrics align with physician preferences. Our research marks the first evidence of LLMs outperforming human experts in clinical text summarization across multiple tasks. This implies that integrating LLMs into clinical workflows could alleviate documentation burden, empowering clinicians to focus more on personalized patient care and the inherently human aspects of medicine.
ABSTRACT
Artificial intelligence (AI) models for automatic generation of narrative radiology reports from images have the potential to enhance efficiency and reduce the workload of radiologists. However, evaluating the correctness of these reports requires metrics that can capture clinically pertinent differences. In this study, we investigate the alignment between automated metrics and radiologists' scoring of errors in report generation. We address the limitations of existing metrics by proposing new metrics, RadGraph F1 and RadCliQ, which demonstrate stronger correlation with radiologists' evaluations. In addition, we analyze the failure modes of the metrics to understand their limitations and provide guidance for metric selection and interpretation. This study establishes RadGraph F1 and RadCliQ as meaningful metrics for guiding future research in radiology report generation.
ABSTRACT
Studies evaluating the local quality of death certification in Brazil focused on completeness of death reporting or inappropriate coding of causes of death, with few investigating missing data. We aimed to use missing and unexpected values in core topics to assess the quality of death certification in Brazilian municipalities, to evaluate its correlation with the percentage of garbage codes, and to employ a data-driven approach with non-linear models to investigate the association of the socioeconomic and health infrastructure context with quality of death statistics among municipalities. This retrospective study used data from the Mortality Information System (2010-2017), and municipal data regarding healthcare infrastructure, socioeconomic characteristics, and death rates. Quality of death certification was assessed by missing or unexpected values in the following core topics: dates of occurrence, registration, and birth, place of occurrence, certifier, sex, and marital status. Models were fit to classify municipalities according to the quality of death certification (poor quality defined as death records with missing or unexpected values in core topics ≥ 80%). Municipalities with poor quality of death certification (43.9%) presented larger populations, lower death rates, lower socioeconomic index, healthcare infrastructure with fewer beds and physicians, and higher proportion of public healthcare facilities. The correlation coefficients between quality of death certification assessed by missing or unexpected values and the proportion of garbage codes were weak (0.11-0.49), but stronger for municipalities with lower socioeconomic scores. The model that best fitted the data was the random forest classifier (ROC AUC = 0.76; precision-recall AUC = 0.78). This innovative way of assessing the quality of death certification could help quality improvement initiatives to include the correctness of essential fields, in addition to garbage coding or completeness of records, especially in municipalities with lower socioeconomic status where garbage coding and the correctness of core topics appear to be related issues.
Subject(s)
Death Certificates , Nonlinear Dynamics , Humans , Brazil , Cities , Retrospective Studies , SeizuresABSTRACT
The Coronavirus disease 2019 (COVID-19) presents open questions in how we clinically diagnose and assess disease course. Recently, chest computed tomography (CT) has shown utility for COVID-19 diagnosis. In this study, we developed Deep COVID DeteCT (DCD), a deep learning convolutional neural network (CNN) that uses the entire chest CT volume to automatically predict COVID-19 (COVID+) from non-COVID-19 (COVID-) pneumonia and normal controls. We discuss training strategies and differences in performance across 13 international institutions and 8 countries. The inclusion of non-China sites in training significantly improved classification performance with area under the curve (AUCs) and accuracies above 0.8 on most test sites. Furthermore, using available follow-up scans, we investigate methods to track patient disease course and predict prognosis.
ABSTRACT
Objective This study describes epidemiological and clinical features of patients with confirmed infection by SARS-CoV-2 diagnosed and treated at Hospital Israelita Albert Einstein , which admitted the first patients with this condition in Brazil. Methods In this retrospective, single-center study, we included all laboratory confirmed COVID-19 cases at Hospital Israelita Albert Einstein , São Paulo, Brazil, from February until March 2020. Demographic, clinical, laboratory and radiological data were analyzed. Results A total of 510 patients with a confirmed diagnosis of COVID-19 were included in this study. Most patients were male (56.9%) with a mean age of 40 years. A history of a close contact with a positive/suspected case was reported by 61.1% of patients and 34.4% had a history of recent international travel. The most common symptoms upon presentation were fever (67.5%), nasal congestion (42.4%), cough (41.6%) and myalgia/arthralgia (36.3%). Chest computed tomography was performed in 78 (15.3%) patients, and 93.6% of those showed abnormal results. Hospitalization was required for 72 (14%) patients and 20 (27.8%) were admitted to the Intensive Care Unit. Regarding clinical treatment, the most often used medicines were intravenous antibiotics (84.7%), chloroquine (45.8%) and oseltamivir (31.9%). Invasive mechanical ventilation was required by 65% of Intensive Care Unit patients. The mean length of stay was 9 days for all patients (22 and 7 days for patients requiring or not intensive care, respectively). Only one patient (1.38%) died during follow-up. Conclusion These results may be relevant for Brazil and other countries with similar characteristics, which are starting to deal with this pandemic.
Subject(s)
Coronavirus Infections/diagnosis , Coronavirus Infections/epidemiology , Pneumonia, Viral/diagnosis , Pneumonia, Viral/epidemiology , Adolescent , Adult , Aged , Betacoronavirus , Brazil , COVID-19 , Child , Child, Preschool , Female , Humans , Infant , Infant, Newborn , Male , Middle Aged , Pandemics , Retrospective Studies , SARS-CoV-2 , Young AdultABSTRACT
Transaxonal degenerations result from neuronal death or the interruption of synaptic connections among neuronal structures. These degenerations are not common but may be recognized by conventional magnetic resonance imaging. OBJECTIVE: The learning objectives of this review include recognition of the imaging characteristics of transaxonal degenerations involving cerebellar connections, the identification of potential encephalic lesions that can lead to these degenerations and correlation of the clinical manifestations with imaging findings that reflect this involvement. METHODS: In this report, we review the neuroanatomical knowledge that provides a basis for identifying potential lesions that can result in these degenerations involving cerebellar structures. RESULTS: Hypertrophic olivary degeneration results from an injury that interrupts any of the components of the Guillain-Mollaret triangle. In this work, we describe cases of lesions in the dentate nucleus and central tegmental tract. The crossed cerebellar diaschisis presents specific imaging findings and clinical correlations associated with its acute and chronic phases. The Wallerian degeneration of the middle cerebellar peduncle is illustrated by fiber injury of the pontine cerebellar tracts. A T2-hyperintensity in the dentate nucleus due to a thalamic acute lesion (in ventral lateral nuclei) is also described. Each condition described here is documented by MRI images and is accompanied by teaching points and an anatomical review of the pathways involved. CONCLUSION: Neurologists and radiologists need to become familiar with the diagnosis of these conditions since their presentations are peculiar and often subtle, and can easily be misdiagnosed as ischemic events, degenerative disease, demyelinating disease or even tumors.
Subject(s)
Cerebellum , Olivary Nucleus , Brain , Magnetic Resonance Imaging , Pons/physiologyABSTRACT
ABSTRACT Transaxonal degenerations result from neuronal death or the interruption of synaptic connections among neuronal structures. These degenerations are not common but may be recognized by conventional magnetic resonance imaging. Objective: The learning objectives of this review include recognition of the imaging characteristics of transaxonal degenerations involving cerebellar connections, the identification of potential encephalic lesions that can lead to these degenerations and correlation of the clinical manifestations with imaging findings that reflect this involvement. Methods: In this report, we review the neuroanatomical knowledge that provides a basis for identifying potential lesions that can result in these degenerations involving cerebellar structures. Results: Hypertrophic olivary degeneration results from an injury that interrupts any of the components of the Guillain-Mollaret triangle. In this work, we describe cases of lesions in the dentate nucleus and central tegmental tract. The crossed cerebellar diaschisis presents specific imaging findings and clinical correlations associated with its acute and chronic phases. The Wallerian degeneration of the middle cerebellar peduncle is illustrated by fiber injury of the pontine cerebellar tracts. A T2-hyperintensity in the dentate nucleus due to a thalamic acute lesion (in ventral lateral nuclei) is also described. Each condition described here is documented by MRI images and is accompanied by teaching points and an anatomical review of the pathways involved. Conclusion: Neurologists and radiologists need to become familiar with the diagnosis of these conditions since their presentations are peculiar and often subtle, and can easily be misdiagnosed as ischemic events, degenerative disease, demyelinating disease or even tumors.
RESUMO Degenerações transaxonais resultam da morte neuronal ou da interrupção de conexões sinápticas entre estruturas neurais. Essas degenerações não são comuns, mas podem ser reconhecidas por imagens de ressonância magnética convencional. Objetivo: Os objetivos de aprendizado desta revisão incluem o reconhecimento das características de imagem de degenerações transaxonais envolvendo conexões cerebelares, a identificação de possíveis lesões encefálicas que podem levar a essas degenerações e a correlação das manifestações clínicas com os achados de imagem que refletem esse envolvimento. Métodos: Neste artigo, revisamos conhecimentos neuroanatômicos que fornecem a base para identificar possíveis lesões que podem resultar nessas degenerações envolvendo estruturas cerebelares. Resultados: A degeneração olivar hipertrófica resulta de uma lesão que interrompe algum dos componentes do triângulo de Guillain-Mollaret. Neste trabalho, descrevemos casos de lesões no núcleo denteado e no trato tegmentar central. A diásquise cerebelar cruzada apresenta achados de imagem específicos e correlações clínicas associadas às suas fases aguda e crônica. A degeneração walleriana dos pedúnculos cerebelares médios é ilustrada pela lesão dos tratos pontino-cerebelares. Uma hiperintensidade em T2 do núcleo denteado devido a uma lesão talâmica aguda (no núcleo ventrolateral) também é descrita. Cada condição aqui descrita é documentada por imagens de ressonância magnética e é acompanhada por pontos didáticos e uma revisão anatômica das vias envolvidas. Conclusão: Neurologistas e radiologistas precisam estar familiarizados com o diagnóstico dessas condições, uma vez que suas apresentações são peculiares e frequentemente sutis, e podem ser facilmente equivocadamente diagnosticadas como lesões isquêmicas, doenças degenerativas, desmielinizantes, ou mesmo tumorais.
Subject(s)
Olivary Nucleus , Cerebellum , Brain , Pons/physiology , Magnetic Resonance ImagingABSTRACT
ABSTRACT Objective This study describes epidemiological and clinical features of patients with confirmed infection by SARS-CoV-2 diagnosed and treated at Hospital Israelita Albert Einstein , which admitted the first patients with this condition in Brazil. Methods In this retrospective, single-center study, we included all laboratory confirmed COVID-19 cases at Hospital Israelita Albert Einstein , São Paulo, Brazil, from February until March 2020. Demographic, clinical, laboratory and radiological data were analyzed. Results A total of 510 patients with a confirmed diagnosis of COVID-19 were included in this study. Most patients were male (56.9%) with a mean age of 40 years. A history of a close contact with a positive/suspected case was reported by 61.1% of patients and 34.4% had a history of recent international travel. The most common symptoms upon presentation were fever (67.5%), nasal congestion (42.4%), cough (41.6%) and myalgia/arthralgia (36.3%). Chest computed tomography was performed in 78 (15.3%) patients, and 93.6% of those showed abnormal results. Hospitalization was required for 72 (14%) patients and 20 (27.8%) were admitted to the Intensive Care Unit. Regarding clinical treatment, the most often used medicines were intravenous antibiotics (84.7%), chloroquine (45.8%) and oseltamivir (31.9%). Invasive mechanical ventilation was required by 65% of Intensive Care Unit patients. The mean length of stay was 9 days for all patients (22 and 7 days for patients requiring or not intensive care, respectively). Only one patient (1.38%) died during follow-up. Conclusion These results may be relevant for Brazil and other countries with similar characteristics, which are starting to deal with this pandemic.
RESUMO Objetivo Descrever as características epidemiológicas e clínicas de pacientes com infecção confirmada pelo SARS-CoV-2, diagnosticados e tratados no Hospital Israelita Albert Einstein, que admitiu os primeiros pacientes com essa condição no Brasil. Métodos Neste estudo retrospectivo, de centro único, incluímos todos os casos com confirmação laboratorial de COVID-19 no Hospital Israelita Albert Einstein, em São Paulo (SP) de fevereiro a março de 2020. Foram analisados dados demográficos, clínicos, laboratoriais e radiológicos. Resultados Foram incluídos 510 pacientes com diagnóstico confirmado de COVID-19. A maioria dos pacientes era do sexo masculino (56,9%), com média de idade de 40 anos. Foi relatada história de contato próximo com um caso positivo/suspeito por 61,1% dos pacientes, e 34,4% tinham história de viagens internacionais recentes. Os sintomas mais comuns foram febre (67,5%), congestão nasal (42,4%), tosse (41,6%) e mialgia/artralgia (36,3%). A tomografia computadorizada de tórax foi realizada em 78 (15,3%) pacientes, e 93,6% deles apresentaram resultados anormais. A hospitalização foi necessária para 72 (14%) pacientes, e 20 (27,8%) foram admitidos na Unidade de Terapia Intensiva. Quanto ao tratamento clínico, os medicamentos mais utilizados foram antibióticos intravenosos (84,7%), cloroquina (45,8%) e oseltamivir (31,9%). A ventilação mecânica invasiva foi necessária em 65% dos pacientes na Unidade de Terapia Intensiva. O tempo médio de internação foi 9 dias para todos os pacientes (22 e 7 dias para pacientes que necessitaram ou não de cuidados intensivos, respectivamente). Apenas um (1,38%) paciente morreu durante o acompanhamento. Conclusão Estes resultados podem ser relevantes para o Brasil e outros países com características semelhantes, que começaram a lidar com essa pandemia.