RESUMO
BACKGROUND AND PURPOSE: Mass effect and vasogenic edema are critical findings on CT of the head. This study compared the accuracy of an artificial intelligence model (Annalise Enterprise CTB) with consensus neuroradiologists' interpretations in detecting mass effect and vasogenic edema. MATERIALS AND METHODS: A retrospective stand-alone performance assessment was conducted on data sets of noncontrast CT head cases acquired between 2016 and 2022 for each finding. The cases were obtained from patients 18 years of age or older from 5 hospitals in the United States. The positive cases were selected consecutively on the basis of the original clinical reports using natural language processing and manual confirmation. The negative cases were selected by taking the next negative case acquired from the same CT scanner after positive cases. Each case was interpreted independently by up-to-three neuroradiologists to establish consensus interpretations. Each case was then interpreted by the artificial intelligence model for the presence of the relevant finding. The neuroradiologists were provided with the entire CT study. The artificial intelligence model separately received thin (≤1.5 mm) and/or thick (>1.5 and ≤5 mm) axial series. RESULTS: The 2 cohorts included 818 cases for mass effect and 310 cases for vasogenic edema. The artificial intelligence model identified mass effect with a sensitivity of 96.6% (95% CI, 94.9%-98.2%) and a specificity of 89.8% (95% CI, 84.7%-94.2%) for the thin series, and 95.3% (95% CI, 93.5%-96.8%) and 93.1% (95% CI, 89.1%-96.6%) for the thick series. It identified vasogenic edema with a sensitivity of 90.2% (95% CI, 82.0%-96.7%) and a specificity of 93.5% (95% CI, 88.9%-97.2%) for the thin series, and 90.0% (95% CI, 84.0%-96.0%) and 95.5% (95% CI, 92.5%-98.0%) for the thick series. The corresponding areas under the curve were at least 0.980. CONCLUSIONS: The assessed artificial intelligence model accurately identified mass effect and vasogenic edema in this CT data set. It could assist the clinical workflow by prioritizing interpretation of cases with abnormal findings, possibly benefiting patients through earlier identification and subsequent treatment.
RESUMO
BACKGROUND AND PURPOSE: The use of MR imaging in emergency settings has been limited by availability, long scan times, and sensitivity to motion. This study assessed the diagnostic performance of an ultrafast brain MR imaging protocol for evaluation of acute intracranial pathology in the emergency department and inpatient settings. MATERIALS AND METHODS: Sixty-six adult patients who underwent brain MR imaging in the emergency department and inpatient settings were included in the study. All patients underwent both the reference and the ultrafast brain MR protocols. Both brain MR imaging protocols consisted of T1-weighted, T2/T2*-weighted, FLAIR, and DWI sequences. The ultrafast MR images were reconstructed by using a machine-learning assisted framework. All images were reviewed by 2 blinded neuroradiologists. RESULTS: The average acquisition time was 2.1 minutes for the ultrafast brain MR protocol and 10 minutes for the reference brain MR protocol. There was 98.5% agreement on the main clinical diagnosis between the 2 protocols. In head-to-head comparison, the reference protocol was preferred in terms of image noise and geometric distortion (P < .05 for both). The ultrafast ms-EPI protocol was preferred over the reference protocol in terms of reduced motion artifacts (P < .01). Overall diagnostic quality was not significantly different between the 2 protocols (P > .05). CONCLUSIONS: The ultrafast brain MR imaging protocol provides high accuracy for evaluating acute pathology while only requiring a fraction of the scan time. Although there was greater image noise and geometric distortion on the ultrafast brain MR protocol images, there was significant reduction in motion artifacts with similar overall diagnostic quality between the 2 protocols.
Assuntos
Encefalopatias , Pacientes Internados , Adulto , Humanos , Imageamento por Ressonância Magnética/métodos , Encéfalo/diagnóstico por imagem , Encéfalo/patologia , Encefalopatias/diagnóstico por imagem , Encefalopatias/patologia , TempoRESUMO
BACKGROUND AND PURPOSE: Deep learning (DL) accelerated MR techniques have emerged as a promising approach to accelerate routine MR exams. While prior studies explored DL acceleration for specific lumbar MRI sequences, a gap remains in comprehending the impact of a fully DL-based MRI protocol on scan time and diagnostic quality for routine lumbar spine MRI. To address this, we assessed the image quality and diagnostic performance of a DL-accelerated lumbar spine MRI protocol in comparison to a conventional protocol. METHODS: We prospectively evaluated 36 consecutive outpatients undergoing non-contrast enhanced lumbar spine MRIs. Both protocols included sagittal T1, T2, STIR, and axial T2-weighted images. Two blinded neuroradiologists independently reviewed images for foraminal stenosis, spinal canal stenosis, nerve root compression, and facet arthropathy. Grading comparison employed the Wilcoxon signed rank test. For the head-to-head comparison, a 5-point Likert scale to assess image quality, considering artifacts, signal-to-noise ratio (SNR), anatomical structure visualization, and overall diagnostic quality. We applied a 15% noninferiority margin to determine whether the DL-accelerated protocol was noninferior. RESULTS: No significant differences existed between protocols when evaluating foraminal and spinal canal stenosis, nerve compression, or facet arthropathy (all p > .05). The DL-spine protocol was noninferior for overall diagnostic quality and visualization of the cord, CSF, intervertebral disc, and nerve roots. However, it exhibited reduced SNR and increased artifact perception. Interobserver reproducibility ranged from moderate to substantial (κ = 0.50-0.76). CONCLUSION: Our study indicates that DL reconstruction in spine imaging effectively reduces acquisition times while maintaining comparable diagnostic quality to conventional MRI.
Assuntos
Aprendizado Profundo , Vértebras Lombares , Imageamento por Ressonância Magnética , Humanos , Imageamento por Ressonância Magnética/métodos , Masculino , Vértebras Lombares/diagnóstico por imagem , Feminino , Estudos Prospectivos , Pessoa de Meia-Idade , Idoso , Razão Sinal-Ruído , Estenose Espinal/diagnóstico por imagem , Adulto , Doenças da Coluna Vertebral/diagnóstico por imagemRESUMO
Flow-related artifacts have been observed in highly accelerated T1-weighted contrast-enhanced wave-controlled aliasing in parallel imaging (CAIPI) magnetization-prepared rapid gradient-echo (MPRAGE) imaging and can lead to diagnostic uncertainty. We developed an optimized flow-mitigated Wave-CAIPI MPRAGE acquisition protocol to reduce these artifacts through testing in a custom-built flow phantom. In the phantom experiment, maximal flow artifact reduction was achieved with the combination of flow compensation gradients and radial reordered k-space acquisition and was included in the optimized sequence. Clinical evaluation of the optimized MPRAGE sequence was performed in 64 adult patients, who all underwent contrast-enhanced Wave-CAIPI MPRAGE imaging without flow-compensation and with optimized flow-compensation parameters. All images were evaluated for the presence of flow-related artifacts, signal-to-noise ratio (SNR), gray-white matter contrast, enhancing lesion contrast, and image sharpness on a 3-point Likert scale. In the 64 cases, the optimized flow mitigation protocol reduced flow-related artifacts in 89% and 94% of the cases for raters 1 and 2, respectively. SNR, gray-white matter contrast, enhancing lesion contrast, and image sharpness were rated as equivalent for standard and flow-mitigated Wave-CAIPI MPRAGE in all subjects. The optimized flow mitigation protocol successfully reduced the presence of flow-related artifacts in the majority of cases.Relevance statementAs accelerated MRI using novel encoding schemes become increasingly adopted in clinical practice, our work highlights the need to recognize and develop strategies to minimize the presence of unexpected artifacts and reduction in image quality as potential compromises to achieving short scan times.Key points⢠Flow-mitigation technique led to an 89-94% decrease in flow-related artifacts.⢠Image quality, signal-to-noise ratio, enhancing lesion conspicuity, and image sharpness were preserved with the flow mitigation technique.⢠Flow mitigation reduced diagnostic uncertainty in cases where flow-related artifacts mimicked enhancing lesions.
Assuntos
Encéfalo , Imageamento por Ressonância Magnética , Adulto , Humanos , Imageamento por Ressonância Magnética/métodos , Razão Sinal-Ruído , Imagens de Fantasmas , ArtefatosRESUMO
In this paper, we present the design and performance of the upgraded University of Florida torsion pendulum facility for testing inertial sensor technology related to space-based gravitational wave observatories and geodesy missions. In particular, much work has been conducted on inertial sensor technology related to the Laser Interferometer Space Antenna (LISA) space gravitational wave observatory mission. A significant upgrade to the facility was the incorporation of a newly designed and fabricated LISA-like gravitational reference sensor (GRS) based on the LISA Pathfinder GRS. Its LISA-like geometry has allowed us to make noise measurements that are more representative of those in LISA and has allowed for the characterization of the mechanisms of noise induced on a LISA GRS and their underlying physics. Noise performance results and experiments exploring the effect of temperature gradients across the sensor will also be discussed. The LISA-like sensor also includes unique UV light injection geometries for UV LED based charge management. Pulsed and DC charge management experiments have been conducted using the University of Florida charge management group's technology readiness level 4 charge management device. These experiments have allowed for the testing of charge management system hardware and techniques as well as characterizations of the dynamics of GRS test mass charging. The work presented here demonstrates the upgraded torsion pendulum's ability to act as an effective testbed for GRS technology.
RESUMO
Machine learning (ML) algorithms to detect critical findings on head CTs may expedite patient management. Most ML algorithms for diagnostic imaging analysis utilize dichotomous classifications to determine whether a specific abnormality is present. However, imaging findings may be indeterminate, and algorithmic inferences may have substantial uncertainty. We incorporated awareness of uncertainty into an ML algorithm that detects intracranial hemorrhage or other urgent intracranial abnormalities and evaluated prospectively identified, 1000 consecutive noncontrast head CTs assigned to Emergency Department Neuroradiology for interpretation. The algorithm classified the scans into high (IC+) and low (IC-) probabilities for intracranial hemorrhage or other urgent abnormalities. All other cases were designated as No Prediction (NP) by the algorithm. The positive predictive value for IC+ cases (N = 103) was 0.91 (CI: 0.84-0.96), and the negative predictive value for IC- cases (N = 729) was 0.94 (0.91-0.96). Admission, neurosurgical intervention, and 30-day mortality rates for IC+ was 75% (63-84), 35% (24-47), and 10% (4-20), compared to 43% (40-47), 4% (3-6), and 3% (2-5) for IC-. There were 168 NP cases, of which 32% had intracranial hemorrhage or other urgent abnormalities, 31% had artifacts and postoperative changes, and 29% had no abnormalities. An ML algorithm incorporating uncertainty classified most head CTs into clinically relevant groups with high predictive values and may help accelerate the management of patients with intracranial hemorrhage or other urgent intracranial abnormalities.
Assuntos
Aprendizado Profundo , Humanos , Incerteza , Tomografia Computadorizada por Raios X/métodos , Hemorragias Intracranianas/diagnóstico por imagem , Algoritmos , Estudos RetrospectivosRESUMO
OBJECTIVE: Platelet-bound complement activation product C4d (PC4d) levels correlate with history of thrombosis in patients with systemic lupus erythematosus (SLE). The present study evaluated whether PC4d levels could assess risk of future thrombosis events. METHODS: PC4d level was measured by flow cytometry. Thromboses were confirmed by electronic medical record data review. RESULTS: The study included 418 patients. Nineteen events (13 arterial and 6 venous) occurred in 15 subjects in the 3 years post-PC4d level measurement. PC4d levels above the optimum cutoff of 13 mean fluorescence intensity (MFI) predicted future arterial thrombosis with a hazard ratio of 4.34 (95% confidence interval [95% CI] 1.03-18.3) (P = 0.046) and a diagnostic odds ratio (OR) of 4.30 (95% CI 1.19-15.54). Negative predictive value of PC4d level of ≤13 MFI for arterial thrombosis was 99% (95% CI 97-100%). Although a PC4d level of >13 MFI did not reach statistical significance for prediction of total thrombosis (arterial and venous) (diagnostics OR 2.50 [95% CI 0.88-7.06]; P = 0.08), it was associated with all thrombosis (n = 70 historic and future arterial and venous events in the 5 years pre- to 3 years post-PC4d level measurement) with an OR of 2.45 (95% CI 1.37-4.32; P = 0.0016). In addition, the negative predictive value of PC4d level of ≤13 MFI for all future thrombosis events was 97% (95% CI 95-99%). CONCLUSIONS: A PC4d level of >13 MFI predicted future arterial thrombosis and was associated with all thrombosis. Patients with SLE presenting with a PC4d level of ≤13 MFI had high probability of not experiencing arterial or any thrombosis in the 3 years afterwards. Taken together, these findings indicate that PC4d levels may help predict the risk of future thrombosis events in SLE.
Assuntos
Lúpus Eritematoso Sistêmico , Trombose , Humanos , Lúpus Eritematoso Sistêmico/complicações , Lúpus Eritematoso Sistêmico/diagnóstico , Trombose/diagnóstico , Trombose/epidemiologia , Trombose/etiologia , Plaquetas , Fatores de RiscoRESUMO
PURPOSE: To develop a robust retrospective motion-correction technique based on repeating k-space guidance lines for improving motion correction in Cartesian 2D and 3D brain MRI. METHODS: The motion guidance lines are inserted into the standard sequence orderings for 2D turbo spin echo and 3D MPRAGE to inform a data consistency-based motion estimation and reconstruction, which can be guided by a low-resolution scout. The extremely limited number of required guidance lines are repeated during each echo train and discarded in the final image reconstruction. Thus, integration within a standard k-space acquisition ordering ensures the expected image quality/contrast and motion sensitivity of that sequence. RESULTS: Through simulation and in vivo 2D multislice and 3D motion experiments, we demonstrate that respectively 2 or 4 optimized motion guidance lines per shot enables accurate motion estimation and correction. Clinically acceptable reconstruction times are achieved through fully separable on-the-fly motion optimizations (Ë1 s/shot) using standard scanner GPU hardware. CONCLUSION: The addition of guidance lines to scout accelerated motion estimation facilitates robust retrospective motion correction that can be effectively introduced without perturbing standard clinical protocols and workflows.
Assuntos
Encéfalo , Imageamento por Ressonância Magnética , Estudos Retrospectivos , Encéfalo/diagnóstico por imagem , Imageamento por Ressonância Magnética/métodos , Movimento (Física) , Simulação por Computador , Imageamento Tridimensional/métodos , Processamento de Imagem Assistida por Computador/métodosRESUMO
INTRODUCTION: Clinical validation studies have demonstrated the ability of accelerated MRI sequences to decrease acquisition time and motion artifact while preserving image quality. The operational benefits, however, have been less explored. Here, we report our initial clinical experience in implementing fast MRI techniques for outpatient brain imaging during the COVID-19 pandemic. METHODS: Aggregate acquisition times were extracted from the medical record on consecutive imaging examinations performed during matched pre-implementation (7/1/2019-12/31/2019) and post-implementation periods (7/1/2020-12/31/2020). Expected acquisition time reduction for each MRI protocol was calculated through manual collection of acquisition times for the conventional and accelerated sequences performed during the pre- and post-implementation periods. Aggregate and expected acquisition times were compared for the five most frequently performed brain MRI protocols: brain without contrast (BR-), brain with and without contrast (BR+), multiple sclerosis (MS), memory loss (MML), and epilepsy (EPL). RESULTS: The expected time reductions for BR-, BR+, MS, MML, and EPL protocols were 6.6 min, 11.9 min, 14 min, 10.8 min, and 14.1 min, respectively. The overall median aggregate acquisition time was 31 [25, 36] min for the pre-implementation period and 18 [15, 22] min for the post-implementation period, with a difference of 13 min (42%). The median acquisition time was reduced by 4 min (25%) for BR-, 14.0 min (44%) for BR+, 14 min (38%) for MS, 11 min (52%) for MML, and 16 min (35%) for EPL. CONCLUSION: The implementation of fast brain MRI sequences significantly reduced the acquisition times for the most commonly performed outpatient brain MRI protocols.
Assuntos
COVID-19 , Esclerose Múltipla , Humanos , Pacientes Ambulatoriais , Pandemias , Imageamento por Ressonância Magnética/métodos , Neuroimagem/métodos , Encéfalo/diagnóstico por imagemRESUMO
OBJECTIVES: High-resolution post-contrast T1-weighted imaging is a workhorse sequence in the evaluation of neurological disorders. The T1-MPRAGE sequence has been widely adopted for the visualization of enhancing pathology in the brain. However, this three-dimensional (3D) acquisition is lengthy and prone to motion artifact, which often compromises diagnostic quality. The goal of this study was to compare a highly accelerated wave-controlled aliasing in parallel imaging (CAIPI) post-contrast 3D T1-MPRAGE sequence (Wave-T1-MPRAGE) with the standard 3D T1-MPRAGE sequence for visualizing enhancing lesions in brain imaging at 3 T. METHODS: This study included 80 patients undergoing contrast-enhanced brain MRI. The participants were scanned with a standard post-contrast T1-MPRAGE sequence (acceleration factor [R] = 2 using GRAPPA parallel imaging technique, acquisition time [TA] = 5 min 18 s) and a prototype post-contrast Wave-T1-MPRAGE sequence (R = 4, TA = 2 min 32 s). Two neuroradiologists performed a head-to-head evaluation of both sequences and rated the visualization of enhancement, sharpness, noise, motion artifacts, and overall diagnostic quality. A 15% noninferiority margin was used to test whether post-contrast Wave-T1-MPRAGE was noninferior to standard T1-MPRAGE. Inter-rater and intra-rater agreement were calculated. Quantitative assessment of CNR/SNR was performed. RESULTS: Wave-T1-MPRAGE was noninferior to standard T1-MPRAGE for delineating enhancing lesions with unanimous agreement in all cases between raters. Wave-T1-MPRAGE was noninferior in the perception of noise (p < 0.001), motion artifact (p < 0.001), and overall diagnostic quality (p < 0.001). CONCLUSION: High-accelerated post-contrast Wave-T1-MPRAGE enabled a two-fold reduction in acquisition time compared to the standard sequence with comparable performance for visualization of enhancing pathology and equivalent perception of noise, motion artifacts and overall diagnostic quality without loss of clinically important information. KEY POINTS: ⢠Post-contrast wave-controlled aliasing in parallel imaging (CAIPI) T1-MPRAGE accelerated the acquisition of three-dimensional (3D) high-resolution post-contrast images by more than two-fold. ⢠Post-contrast Wave-T1-MPRAGE was noninferior to standard T1-MPRAGE with unanimous agreement between reviewers (100% in 80 cases) for the visualization of intracranial enhancing lesions. ⢠Wave-T1-MPRAGE was equivalent to the standard sequence in the perception of noise in 94% (75 of 80) of cases and was preferred in 16% (13 of 80) of cases for decreased motion artifact.
Assuntos
Imageamento Tridimensional , Imageamento por Ressonância Magnética , Humanos , Imageamento por Ressonância Magnética/métodos , Imageamento Tridimensional/métodos , Encéfalo/diagnóstico por imagem , Artefatos , Movimento (Física)RESUMO
There is increasing interest in deep UV Light-Emitting Diodes (LEDs) for applications in water purification, virus inactivation, sterilization, bioagent detection, and UV curing, as well as charge management control in the Laser Interferometer Space Antenna (LISA), which will be the first gravitational wave detector in space. To fully understand the current state of commercial UV LEDs and assess their performance for use on LISA, large numbers of UV LEDs need to be tested across a range of temperatures while operating in air or in a vacuum. We describe a new hardware system designed to accommodate a high volume of UV LED performance tests and present the performance testing results from over 200 UV LEDs with wavelengths in the 250 nm range.
Assuntos
Esterilização , Inativação de Vírus , Temperatura , VácuoRESUMO
Risk prediction requires comprehensive integration of clinical information and concurrent radiological findings. We present an upgraded chest radiograph (CXR) explainable artificial intelligence (xAI) model, which was trained on 241,723 well-annotated CXRs obtained prior to the onset of the COVID-19 pandemic. Mean area under the receiver operating characteristic curve (AUROC) for detection of 20 radiographic features was 0.955 (95% CI 0.938-0.955) on PA view and 0.909 (95% CI 0.890-0.925) on AP view. Coexistent and correlated radiographic findings are displayed in an interpretation table, and calibrated classifier confidence is displayed on an AI scoreboard. Retrieval of similar feature patches and comparable CXRs from a Model-Derived Atlas provides justification for model predictions. To demonstrate the feasibility of a fine-tuning approach for efficient and scalable development of xAI risk prediction models, we applied our CXR xAI model, in combination with clinical information, to predict oxygen requirement in COVID-19 patients. Prediction accuracy for high flow oxygen (HFO) and mechanical ventilation (MV) was 0.953 and 0.934 at 24 h and 0.932 and 0.836 at 72 h from the time of emergency department (ED) admission, respectively. Our CXR xAI model is auditable and captures key pathophysiological manifestations of cardiorespiratory diseases and cardiothoracic comorbidities. This model can be efficiently and broadly applied via a fine-tuning approach to provide fully automated risk and outcome predictions in various clinical scenarios in real-world practice.
Assuntos
COVID-19 , Oxigênio , Humanos , COVID-19/diagnóstico por imagem , Inteligência Artificial , Pandemias , PacientesRESUMO
Fast MRI and portable MRI are emerging as promising technologies to improve the speed, efficiency, and availability of MR imaging. Fast MRI methods are increasingly being adopted to create screening protocols for the diagnosis and management of acute pathology in the emergency department. Faster imaging can facilitate timely diagnosis, reduce motion artifacts, and improve departmental MR operations. Point-of-care and portable MRI are emerging technologies that require radiologists to reenvision the role of MRI as a tool with greater accessibility, fewer siting constraints, and the ability to provide valuable diagnostic information at the bedside. Recently introduced commercially available pulse sequences and new MRI scanners are bringing these technologies closer to the patient's clinical setting, and we expect their use to only increase over the coming decade. This article provides an overview of these emerging technologies for emergency radiologists.
Assuntos
Artefatos , Imageamento por Ressonância Magnética , Humanos , Imageamento por Ressonância Magnética/métodosRESUMO
OBJECTIVES: Wave-CAIPI (Controlled Aliasing in Parallel Imaging) enables dramatic reduction in acquisition time of 3D MRI sequences such as 3D susceptibility-weighted imaging (SWI) but has not been clinically evaluated at 1.5 T. We sought to compare highly accelerated Wave-CAIPI SWI (Wave-SWI) with two alternative standard sequences, conventional three-dimensional SWI and two-dimensional T2*-weighted Gradient-Echo (T2*w-GRE), in patients undergoing routine brain MRI at 1.5 T. METHODS: In this study, 172 patients undergoing 1.5 T brain MRI were scanned with a more commonly used susceptibility sequence (standard SWI or T2*w-GRE) and a highly accelerated Wave-SWI sequence. Two radiologists blinded to the acquisition technique scored each sequence for visualization of pathology, motion and signal dropout artifacts, image noise, visualization of normal anatomy (vessels and basal ganglia mineralization), and overall diagnostic quality. Superiority testing was performed to compare Wave-SWI to T2*w-GRE, and non-inferiority testing with 15% margin was performed to compare Wave-SWI to standard SWI. RESULTS: Wave-SWI performed superior in terms of visualization of pathology, signal dropout artifacts, visualization of normal anatomy, and overall image quality when compared to T2*w-GRE (all p < 0.001). Wave-SWI was non-inferior to standard SWI for visualization of normal anatomy and pathology, signal dropout artifacts, and overall image quality (all p < 0.001). Wave-SWI was superior to standard SWI for motion artifact (p < 0.001), while both conventional susceptibility sequences were superior to Wave-SWI for image noise (p < 0.001). CONCLUSIONS: Wave-SWI can be performed in a 1.5 T clinical setting with robust performance and preservation of diagnostic quality. KEY POINTS: ⢠Wave-SWI accelerated the acquisition of 3D high-resolution susceptibility images in 70% of the acquisition time of the conventional T2*GRE. ⢠Wave-SWI performed superior to T2*w-GRE for visualization of pathology, signal dropout artifacts, and overall diagnostic image quality. ⢠Wave-SWI was noninferior to standard SWI for visualization of normal anatomy and pathology, signal dropout artifacts, and overall diagnostic image quality.
Assuntos
Imageamento por Ressonância Magnética , Neuroimagem , Artefatos , Encéfalo/diagnóstico por imagem , Humanos , Imageamento Tridimensional/métodos , Imageamento por Ressonância Magnética/métodosAssuntos
Doenças de Pequenos Vasos Cerebrais , Encéfalo/diagnóstico por imagem , Encéfalo/patologia , Doenças de Pequenos Vasos Cerebrais/complicações , Doenças de Pequenos Vasos Cerebrais/diagnóstico por imagem , Doenças de Pequenos Vasos Cerebrais/patologia , Humanos , Imageamento por Ressonância Magnética/métodos , Espectroscopia de Ressonância MagnéticaRESUMO
The inability to accurately, efficiently label large, open-access medical imaging datasets limits the widespread implementation of artificial intelligence models in healthcare. There have been few attempts, however, to automate the annotation of such public databases; one approach, for example, focused on labor-intensive, manual labeling of subsets of these datasets to be used to train new models. In this study, we describe a method for standardized, automated labeling based on similarity to a previously validated, explainable AI (xAI) model-derived-atlas, for which the user can specify a quantitative threshold for a desired level of accuracy (the probability-of-similarity, pSim metric). We show that our xAI model, by calculating the pSim values for each clinical output label based on comparison to its training-set derived reference atlas, can automatically label the external datasets to a user-selected, high level of accuracy, equaling or exceeding that of human experts. We additionally show that, by fine-tuning the original model using the automatically labelled exams for retraining, performance can be preserved or improved, resulting in a highly accurate, more generalized model.