1
|
Differential associations between abnormal cardiac left ventricular geometry types and cerebral white matter disease. J Stroke Cerebrovasc Dis 2024; 33:107709. [PMID: 38570059 DOI: 10.1016/j.jstrokecerebrovasdis.2024.107709] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/07/2023] [Revised: 02/19/2024] [Accepted: 04/01/2024] [Indexed: 04/05/2024] Open
Abstract
OBJECTIVES Reduced cardiac outflow due to left ventricular hypertrophy has been suggested as a potential risk factor for development of cerebral white matter disease. Our study aimed to examine the correlation between left ventricular geometry and white matter disease volume to establish a clearer understanding of their relationship, as it is currently not well-established. METHODS Consecutive patients from 2016 to 2021 who were ≥18 years and underwent echocardiography, cardiac MRI, and brain MRI within one year were included. Four categories of left ventricular geometry were defined based on left ventricular mass index and relative wall thickness on echocardiography. White matter disease volume was quantified using an automated algorithm applied to axial T2 FLAIR images and compared across left ventricular geometry categories. RESULTS We identified 112 patients of which 34.8 % had normal left ventricular geometry, 20.5 % had eccentric hypertrophy, 21.4 % had concentric remodeling, and 23.2 % had concentric hypertrophy. White matter disease volume was highest in patients with concentric hypertrophy and concentric remodeling, compared to eccentric hypertrophy and normal morphology with a trend-P value of 0.028. Patients with higher relative wall thickness had higher white matter disease volume (10.73 ± 10.29 cc vs 5.89 ± 6.46 cc, P = 0.003), compared to those with normal relative wall thickness. CONCLUSION Our results showed that abnormal left ventricular geometry is associated with higher white matter disease burden, particularly among those with abnormal relative wall thickness. Future studies are needed to explore causative relationships and potential therapeutic options that may mediate the adverse left ventricular remodeling and its effect in slowing white matter disease progression.
Collapse
|
2
|
Fair evaluation of federated learning algorithms for automated breast density classification: The results of the 2022 ACR-NCI-NVIDIA federated learning challenge. Med Image Anal 2024; 95:103206. [PMID: 38776844 DOI: 10.1016/j.media.2024.103206] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 06/22/2023] [Revised: 02/15/2024] [Accepted: 05/13/2024] [Indexed: 05/25/2024]
Abstract
The correct interpretation of breast density is important in the assessment of breast cancer risk. AI has been shown capable of accurately predicting breast density, however, due to the differences in imaging characteristics across mammography systems, models built using data from one system do not generalize well to other systems. Though federated learning (FL) has emerged as a way to improve the generalizability of AI without the need to share data, the best way to preserve features from all training data during FL is an active area of research. To explore FL methodology, the breast density classification FL challenge was hosted in partnership with the American College of Radiology, Harvard Medical Schools' Mass General Brigham, University of Colorado, NVIDIA, and the National Institutes of Health National Cancer Institute. Challenge participants were able to submit docker containers capable of implementing FL on three simulated medical facilities, each containing a unique large mammography dataset. The breast density FL challenge ran from June 15 to September 5, 2022, attracting seven finalists from around the world. The winning FL submission reached a linear kappa score of 0.653 on the challenge test data and 0.413 on an external testing dataset, scoring comparably to a model trained on the same data in a central location.
Collapse
|
3
|
Greater burden of white matter lesions and silent infarcts ipsilateral to carotid stenosis. J Stroke Cerebrovasc Dis 2023; 32:107287. [PMID: 37531723 DOI: 10.1016/j.jstrokecerebrovasdis.2023.107287] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/05/2023] [Revised: 07/25/2023] [Accepted: 07/29/2023] [Indexed: 08/04/2023] Open
Abstract
OBJECTIVES Carotid stenosis may cause silent cerebrovascular disease (CVD) through atheroembolism and hypoperfusion. If so, revascularization may slow progression of silent CVD. We aimed to compare the presence and severity of silent CVD to the degree of carotid bifurcation stenosis by cerebral hemisphere. MATERIALS AND METHODS Patients age ≥40 years with carotid stenosis >50% by carotid ultrasound who underwent MRI brain from 2011-2015 at Mayo Clinic were included. Severity of carotid stenosis was classified by carotid duplex ultrasound as 50-69% (moderate), 70-99% (severe), or occluded. White matter lesion (WML) volume was quantified using an automated deep-learning algorithm applied to axial T2 FLAIR images. Differences in WML volume and prevalent silent infarcts were compared across hemispheres and severity of carotid stenosis. RESULTS Of the 183 patients, mean age was 71±10 years, and 39.3% were female. Moderate stenosis was present in 35.5%, severe stenosis in 46.5% and occlusion in 18.0%. Patients with carotid stenosis had greater WML volume ipsilateral to the side of carotid stenosis than the contralateral side (mean difference, 0.42±0.21cc, p=0.046). Higher degrees of stenosis were associated with greater hemispheric difference in WML volume (moderate vs. severe; 0.16±0.27cc vs 0.74±0.31cc, p=0.009). Prevalence of silent infarct was 23.5% and was greater on the side of carotid stenosis than the contralateral side (hemispheric difference 8.8%±3.2%, p=0.006). Higher degrees of stenosis were associated with higher burden of silent infarcts (moderate vs severe, 10.8% vs 31.8%; p=0.002). CONCLUSIONS WML and silent infarcts were greater on the side of severe carotid stenosis.
Collapse
|
4
|
Pre-deployment assessment of an AI model to assist radiologists in chest X-ray detection and identification of lead-less implanted electronic devices for pre-MRI safety screening: realized implementation needs and proposed operational solutions. J Med Imaging (Bellingham) 2022; 9:054504. [PMID: 36310648 PMCID: PMC9603740 DOI: 10.1117/1.jmi.9.5.054504] [Citation(s) in RCA: 1] [Impact Index Per Article: 0.5] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/10/2021] [Accepted: 09/23/2022] [Indexed: 09/29/2023] Open
Abstract
Purpose Chest X-ray (CXR) use in pre-MRI safety screening, such as for lead-less implanted electronic device (LLIED) recognition, is common. To assist CXR interpretation, we "pre-deployed" an artificial intelligence (AI) model to assess (1) accuracies in LLIED-type (and consequently safety-level) identification, (2) safety implications of LLIED nondetections or misidentifications, (3) infrastructural or workflow requirements, and (4) demands related to model adaptation to real-world conditions. Approach A two-tier cascading methodology for LLIED detection/localization and identification on a frontal CXR was applied to evaluate the performance of the original nine-class AI model. With the unexpected early appearance of LLIED types during simulated real-world trialing, retraining of a newer 12-class version preceded retrialing. A zero footprint (ZF) graphical user interface (GUI)/viewer with DICOM-based output was developed for inference-result display and adjudication, supporting end-user engagement and model continuous learning and/or modernization. Results During model testing or trialing using both the nine-class and 12-class models, robust detection/localization was consistently 100%, with mAP 0.99 from fivefold cross-validation. Safety-level categorization was high during both testing ( AUC ≥ 0.98 and ≥ 0.99 , respectively) and trialing (accuracy 98% and 97%, respectively). LLIED-type identifications by the two models during testing (1) were 98.9% and 99.5% overall correct and (2) consistently showed AUC ≥ 0.92 (1.00 for 8/9 and 9/12 LLIED-types, respectively). Pre-deployment trialing of both models demonstrated overall type-identification accuracies of 94.5% and 95%, respectively. Of the small number of misidentifications, none involved MRI-stringently conditional or MRI-unsafe types of LLIEDs. Optimized ZF GUI/viewer operations led to greater user-friendliness for radiologist engagement. Conclusions Our LLIED-related AI methodology supports (1) 100% detection sensitivity, (2) high identification (including MRI-safety) accuracy, and (3) future model deployment with facilitated inference-result display and adjudication for ongoing model adaptation to future real-world experiences.
Collapse
|
5
|
Abstract TMP81: The Relationship Between Carotid Stenosis And Silent Cerebrovascular Disease. Stroke 2022. [DOI: 10.1161/str.53.suppl_1.tmp81] [Citation(s) in RCA: 0] [Impact Index Per Article: 0] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/16/2022]
Abstract
Introduction:
Patients with carotid stenosis may have silent cerebrovascular disease due to chronic hypoperfusion and atheroembolism. If so, progression of silent disease may be preventable through revascularization. We aim to evaluate the association between carotid stenosis/occlusion and cerebral small vessel disease (SVD) burden ipsilateral to the cervical carotid stenosis.
Methods:
Patients age ≥40 years with carotid stenosis or occlusion on carotid ultrasound who underwent MRI brain and intracranial angiogram with gradable SVD on MRI from 2011-2015 were included. Severity of carotid stenosis was defined using NASCET criteria as 50-69% (moderate), 70-99% (high-grade), or occluded. WMH volume was quantified using an automated artificial intelligence algorithm applied to axial T2 FLAIR images. Images were also scored for presence of chronic lacune, cerebral microbleeds (CMB), and silent infarcts using STRIVE criteria. Differences in WMD volume between hemispheres ipsilateral and contralateral to the carotid stenosis were calculated, and prevalence of SVD and silent infarct were compared across severities of carotid stenosis.
Results:
Of the 183 patients, mean age was 71±10 years and 38% were female. Moderate and severe stenosis was present in 36% and 46.5%, respectively, and 18% had carotid occlusion. Mean WMH volume was 7271±522mm
3
. Patients with carotid stenosis had greater WMH volume ipsilateral to the side of carotid stenosis than the contralateral side (mean difference 224±206mm
3
), and higher severity of stenosis was associated with greater hemispheric difference in WMH volume (moderate vs high; 164±274mm
3
vs 743±309 mm
3
, p=0.009). Silent infarcts were more prevalent in high-grade carotid stenosis than moderate grade (moderate vs high, 10.8% vs 36.5%; p=0.002). There were no differences in either lacunae or CMBs across severities.
Conclusions:
Greater than one third of patients with high-grade carotid stenosis had ipsilateral silent brain infarction, and the volume of WMH was greater on the side of carotid stenosis. Future analyses should consider the potential yield of screening for carotid atherosclerosis in patients with varying degrees of hemispheric WMH asymmetry.
Collapse
|
6
|
Are quantitative features of lung nodules reproducible at different CT acquisition and reconstruction parameters? PLoS One 2020; 15:e0240184. [PMID: 33057454 PMCID: PMC7561205 DOI: 10.1371/journal.pone.0240184] [Citation(s) in RCA: 9] [Impact Index Per Article: 2.3] [Reference Citation Analysis] [Abstract] [MESH Headings] [Grants] [Track Full Text] [Download PDF] [Figures] [Journal Information] [Subscribe] [Scholar Register] [Received: 05/13/2020] [Accepted: 09/22/2020] [Indexed: 12/30/2022] Open
Abstract
Consistency and duplicability in Computed Tomography (CT) output is essential to quantitative imaging for lung cancer detection and monitoring. This study of CT-detected lung nodules investigated the reproducibility of volume-, density-, and texture-based features (outcome variables) over routine ranges of radiation dose, reconstruction kernel, and slice thickness. CT raw data of 23 nodules were reconstructed using 320 acquisition/reconstruction conditions (combinations of 4 doses, 10 kernels, and 8 thicknesses). Scans at 12.5%, 25%, and 50% of protocol dose were simulated; reduced-dose and full-dose data were reconstructed using conventional filtered back-projection and iterative-reconstruction kernels at a range of thicknesses (0.6-5.0 mm). Full-dose/B50f kernel reconstructions underwent expert segmentation for reference Region-Of-Interest (ROI) and nodule volume per thickness; each ROI was applied to 40 corresponding images (combinations of 4 doses and 10 kernels). Typical texture analysis metrics (including 5 histogram features, 13 Gray Level Co-occurrence Matrix, 5 Run Length Matrix, 2 Neighboring Gray-Level Dependence Matrix, and 3 Neighborhood Gray-Tone Difference Matrix) were computed per ROI. Reconstruction conditions resulting in no significant change in volume, density, or texture metrics were identified as "compatible pairs" for a given outcome variable. Our results indicate that as thickness increases, volumetric reproducibility decreases, while reproducibility of histogram- and texture-based features across different acquisition and reconstruction parameters improves. To achieve concomitant reproducibility of volumetric and radiomic results across studies, balanced standardization of the imaging acquisition parameters is required.
Collapse
|
7
|
Automated Brain Metastases Detection Framework for T1-Weighted Contrast-Enhanced 3D MRI. IEEE J Biomed Health Inform 2020; 24:2883-2893. [DOI: 10.1109/jbhi.2020.2982103] [Citation(s) in RCA: 28] [Impact Index Per Article: 7.0] [Reference Citation Analysis] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 11/06/2022]
|
8
|
Bone grafting history affects soft tissue healing following implant placement. J Periodontol 2020; 92:234-243. [PMID: 32779206 DOI: 10.1002/jper.19-0709] [Citation(s) in RCA: 5] [Impact Index Per Article: 1.3] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/10/2019] [Revised: 05/06/2020] [Accepted: 05/06/2020] [Indexed: 11/06/2022]
Abstract
BACKGROUND This study aimed to determine and compare soft tissue healing outcomes following implant placement in grafted (GG) and non-grafted bone (NGG). METHODS Patients receiving single implant in a tooth-bound maxillary non-molar site were recruited. Clinical healing was documented. Volume and content of wound fluid (WF; at 3, 6, and 9 days) were compared with adjacent gingival crevicular fluid (GCF; at baseline, 1, and 4 months). Buccal flap blood perfusion recovery and changes in bone thickness were recorded. Linear mixed model regression analysis and generalized estimating equations with Bonferroni adjustments were conducted for repeated measures. RESULTS Twenty-five patients (49 ± 4 years; 13 males; nine NGG) completed the study. Soft tissue closure was slower in GG (P < 0.01). Differential response in WF/GCF protein concentrations was detected for ACTH (increased in GG only) and insulin, leptin, osteocalcin (decreased in NGG only) at day 6 (P ≤0.04), with no inter-group differences at any time(P > 0.05). Blood perfusion rate decreased immediately postoperatively (P < 0.01, GG) followed by 3-day hyperemia (P > 0.05 both groups). The recovery to baseline values was almost complete for NGG whereas GG stayed ischemic even at 4 months (P = 0.05). Buccal bone thickness changes were significant in GG sites (P ≤ 0.05). CONCLUSION History of bone grafting alters the clinical, physiological, and molecular healing response of overlying soft tissues after implant placement surgery.
Collapse
|
9
|
Automated coronary artery atherosclerosis detection and weakly supervised localization on coronary CT angiography with a deep 3-dimensional convolutional neural network. Comput Med Imaging Graph 2020; 83:101721. [PMID: 32470854 DOI: 10.1016/j.compmedimag.2020.101721] [Citation(s) in RCA: 22] [Impact Index Per Article: 5.5] [Reference Citation Analysis] [Abstract] [Key Words] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 11/27/2019] [Revised: 03/09/2020] [Accepted: 03/30/2020] [Indexed: 11/26/2022]
Abstract
We propose a fully automated algorithm based on a deep learning framework enabling screening of a coronary computed tomography angiography (CCTA) examination for confident detection of the presence or absence of coronary artery atherosclerosis. The system starts with extracting the coronary arteries and their branches from CCTA datasets and representing them with multi-planar reformatted volumes; pre-processing and augmentation techniques are then applied to increase the robustness and generalization ability of the system. A 3-dimensional convolutional neural network (3D-CNN) is utilized to model pathological changes (e.g., atherosclerotic plaques) in coronary vessels. The system learns the discriminatory features between vessels with and without atherosclerosis. The discriminative features at the final convolutional layer are visualized with a saliency map approach to provide visual clues related to atherosclerosis likelihood and location. We have evaluated the system on a reference dataset representing 247 patients with atherosclerosis and 246 patients free of atherosclerosis. With five fold cross-validation, an Accuracy = 90.9%, Positive Predictive Value = 58.8%, Sensitivity = 68.9%, Specificity of 93.6%, and Negative Predictive Value (NPV) = 96.1% are achieved at the artery/branch level with threshold 0.5. The average area under the receiver operating characteristic curve is 0.91. The system indicates a high NPV, which may be potentially useful for assisting interpreting physicians in excluding coronary atherosclerosis in patients with acute chest pain.
Collapse
|
10
|
Performance of a Deep Neural Network Algorithm Based on a Small Medical Image Dataset: Incremental Impact of 3D-to-2D Reformation Combined with Novel Data Augmentation, Photometric Conversion, or Transfer Learning. J Digit Imaging 2020; 33:431-438. [PMID: 31625028 PMCID: PMC7165215 DOI: 10.1007/s10278-019-00267-3] [Citation(s) in RCA: 6] [Impact Index Per Article: 1.5] [Reference Citation Analysis] [Abstract] [Key Words] [MESH Headings] [Grants] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/05/2023] Open
Abstract
Collecting and curating large medical-image datasets for deep neural network (DNN) algorithm development is typically difficult and resource-intensive. While transfer learning (TL) decreases reliance on large data collections, current TL implementations are tailored to two-dimensional (2D) datasets, limiting applicability to volumetric imaging (e.g., computed tomography). Targeting performance enhancement of a DNN algorithm based on a small image dataset, we assessed incremental impact of 3D-to-2D projection methods, one supporting novel data augmentation (DA); photometric grayscale-to-color conversion (GCC); and/or TL on training of an algorithm from a small coronary computed tomography angiography (CCTA) dataset (200 examinations, 50% with atherosclerosis and 50% atherosclerosis-free) producing 245 diseased and 1127 normal coronary arteries/branches. Volumetric CCTA data was converted to a 2D format creating both an Aggregate Projection View (APV) and a Mosaic Projection View (MPV), supporting DA per vessel; both grayscale and color-mapped versions of each view were also obtained. Training was performed both without and with TL, and algorithm performance of all permutations was compared using area under the receiver operating characteristics curve. Without TL, APV performance was 0.74 and 0.87 on grayscale and color images, respectively, compared to 0.90 and 0.87 for MPV. With TL, APV performance was 0.78 and 0.88 on grayscale and color images, respectively, compared with 0.93 and 0.91 for MPV. In conclusion, TL enhances performance of a DNN algorithm from a small volumetric dataset after proposed 3D-to-2D reformatting, but additive gain is achieved with application of either GCC to APV or the proposed novel MPV technique for DA.
Collapse
|
11
|
Detection and localisation of hip fractures on anteroposterior radiographs with artificial intelligence: proof of concept. Clin Radiol 2019; 75:237.e1-237.e9. [PMID: 31787211 DOI: 10.1016/j.crad.2019.10.022] [Citation(s) in RCA: 23] [Impact Index Per Article: 4.6] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 08/22/2019] [Accepted: 10/30/2019] [Indexed: 11/29/2022]
Abstract
AIM To investigate the feasibility of applying a deep convolutional neural network (CNN) for detection/localisation of acute proximal femoral fractures (APFFs) on hip radiographs. MATERIALS AND METHODS This study had institutional review board approval. Radiographs of 307 patients with APFFs and 310 normal patients were identified. A split ratio of 3/1/1 was used to create training, validation, and test datasets. To test the validity of the proposed model, a 20-fold cross-validation was performed. The anonymised images from the test cohort were shown to two groups of radiologists: musculoskeletal radiologists and diagnostic radiology residents. Each reader was asked to assess if there was a fracture and localise it if one was detected. The area under the receiver operator characteristics curve (AUC), sensitivity, and specificity were calculated for the CNN and readers. RESULTS The mean AUC was 0.9944 with a standard deviation of 0.0036. Mean sensitivity and specificity for fracture detection was 97.1% (81.5/84) and 96.7% (118/122), respectively. There was good concordance with saliency maps for lesion identification, but sensitivity was lower for characterising location (subcapital/transcervical, 84.1%; basicervical/intertrochanteric, 77%; subtrochanteric, 20%). Musculoskeletal radiologists showed a sensitivity and specificity for fracture detection of 100% and 100% respectively, while residents showed 100% and 96.8%, respectively. For fracture localisation, the performance decreased slightly for human readers. CONCLUSION The proposed CNN algorithm showed high accuracy for detection of APFFs, but the performance was lower for fracture localisation. Overall performance of the CNN was lower than that of radiologists, especially in localizing fracture location.
Collapse
|
12
|
A User Interface for Optimizing Radiologist Engagement in Image Data Curation for Artificial Intelligence. Radiol Artif Intell 2019; 1:e180095. [PMID: 33937804 DOI: 10.1148/ryai.2019180095] [Citation(s) in RCA: 17] [Impact Index Per Article: 3.4] [Reference Citation Analysis] [Abstract] [Track Full Text] [Journal Information] [Subscribe] [Scholar Register] [Received: 12/28/2018] [Revised: 06/14/2019] [Accepted: 06/25/2019] [Indexed: 11/11/2022]
Abstract
Purpose To delineate image data curation needs and describe a locally designed graphical user interface (GUI) to aid radiologists in image annotation for artificial intelligence (AI) applications in medical imaging. Materials and Methods GUI components support image analysis toolboxes, picture archiving and communication system integration, third-party applications, processing of scripting languages, and integration of deep learning libraries. For clinical AI applications, GUI components included two-dimensional segmentation and classification; three-dimensional segmentation and quantification; and three-dimensional segmentation, quantification, and classification. To assess radiologist engagement and performance efficiency associated with GUI-related capabilities, image annotation rate (studies per day) and speed (minutes per case) were evaluated in two clinical scenarios of varying complexity: hip fracture detection and coronary atherosclerotic plaque demarcation and stenosis grading. Results For hip fracture, 1050 radiographs were annotated over 7 days (150 studies per day; median speed: 10 seconds per study [interquartile range, 3-21 seconds per study]). A total of 294 coronary CT angiographic studies with 1843 arteries and branches were annotated for atherosclerotic plaque over 23 days (15.2 studies [80.1 vessels] per day; median speed: 6.08 minutes per study [interquartile range, 2.8-10.6 minutes per study] and 73 seconds per vessel [interquartile range, 20.9-155 seconds per vessel]). Conclusion GUI-component compatibility with common image analysis tools facilitates radiologist engagement in image data curation, including image annotation, supporting AI application development and evolution for medical imaging. When complemented by other GUI elements, a continuous integrated workflow supporting formation of an agile deep neural network life cycle results.Supplemental material is available for this article.© RSNA, 2019.
Collapse
|
13
|
Abstract
Radiology and Enterprise Medical Imaging Extensions (REMIX) is a platform originally designed to both support the medical imaging-driven clinical and clinical research operational needs of Department of Radiology of The Ohio State University Wexner Medical Center. REMIX accommodates the storage and handling of “big imaging data,” as needed for large multi-disciplinary cancer-focused programs. The evolving REMIX platform contains an array of integrated tools/software packages for the following: (1) server and storage management; (2) image reconstruction; (3) digital pathology; (4) de-identification; (5) business intelligence; (6) texture analysis; and (7) artificial intelligence. These capabilities, along with documentation and guidance, explaining how to interact with a commercial system (e.g., PACS, EHR, commercial database) that currently exists in clinical environments, are to be made freely available.
Collapse
|
14
|
Automated Critical Test Findings Identification and Online Notification System Using Artificial Intelligence in Imaging. Radiology 2017; 285:923-931. [PMID: 28678669 DOI: 10.1148/radiol.2017162664] [Citation(s) in RCA: 147] [Impact Index Per Article: 21.0] [Reference Citation Analysis] [Abstract] [Journal Information] [Subscribe] [Scholar Register] [Indexed: 02/01/2023]
Abstract
Purpose To evaluate the performance of an artificial intelligence (AI) tool using a deep learning algorithm for detecting hemorrhage, mass effect, or hydrocephalus (HMH) at non-contrast material-enhanced head computed tomographic (CT) examinations and to determine algorithm performance for detection of suspected acute infarct (SAI). Materials and Methods This HIPAA-compliant retrospective study was completed after institutional review board approval. A training and validation dataset of noncontrast-enhanced head CT examinations that comprised 100 examinations of HMH, 22 of SAI, and 124 of noncritical findings was obtained resulting in 2583 representative images. Examinations were processed by using a convolutional neural network (deep learning) using two different window and level configurations (brain window and stroke window). AI algorithm performance was tested on a separate dataset containing 50 examinations with HMH findings, 15 with SAI findings, and 35 with noncritical findings. Results Final algorithm performance for HMH showed 90% (45 of 50) sensitivity (95% confidence interval [CI]: 78%, 97%) and 85% (68 of 80) specificity (95% CI: 76%, 92%), with area under the receiver operating characteristic curve (AUC) of 0.91 with the brain window. For SAI, the best performance was achieved with the stroke window showing 62% (13 of 21) sensitivity (95% CI: 38%, 82%) and 96% (27 of 28) specificity (95% CI: 82%, 100%), with AUC of 0.81. Conclusion AI using deep learning demonstrates promise for detecting critical findings at noncontrast-enhanced head CT. A dedicated algorithm was required to detect SAI. Detection of SAI showed lower sensitivity in comparison to detection of HMH, but showed reasonable performance. Findings support further investigation of the algorithm in a controlled and prospective clinical setting to determine whether it can independently screen noncontrast-enhanced head CT examinations and notify the interpreting radiologist of critical findings. © RSNA, 2017 Online supplemental material is available for this article.
Collapse
|