Search

BackgroundStroke is a leading cause of death and disability. The ability to quickly identify the presence of acute infarct and quantify the volume on magnetic resonance imaging (MRI) has important treatment implications. MethodsWe developed a machine learning model that used the apparent diffusion coefficient and diffusion weighted imaging series. It was trained on 6,657 MRI studies. All studies were labelled positive or negative for infarct (classification annotation) with 377 having the region of interest outlined (segmentation annotation). The different annotation types facilitated training on more studies while not requiring the extensive time to manually segment every study. We initially validated the model on studies sequestered from the training set. We then tested the model on studies from three clinical scenarios: consecutive stroke team activations for 6-months at the hospital that provided training data, consecutive stroke team activations for 6-months at a hospital that did not provide training data, and an international site. The model results were compared to radiologist ground truth interpretations.ResultsThe model performed better when trained on classification and segmentation annotations (area under the receiver operating curve [AUROC] 0.995 [95% CI, 0.992-0.998] and median Dice coefficient for segmentation overlap of 0.797 [IQR, 0.642-0.861]) compared to segmentation annotations alone (AUROC 0.982 [95% CI, 0.972-0.990] and Dice coefficient 0.776 [IQR, 0.584-0.857]). The model accurately identified infarcts for training hospital stroke team activations (AUROC 0.964 [95% CI, 0.943-0.982], 381 studies), non-training hospital stroke team activations (AUROC 0.981 [95% CI, 0.966-0.993], 247 studies), and at the international site (AUROC 0.998 [95% CI, 0.993-1.000], 171 studies). The model accurately segmented infarcts with Pearson correlation comparing model output and ground truth volumes between 0.968-0.986 for the three scenarios.ConclusionsAcute infarct can be accurately detected and segmented on MRI in real-world clinical scenarios using a machine learning model.

PurposeMotion-impaired CT images can result in limited or suboptimal diagnostic interpretation (with missed or miscalled lesions) and patient recall. We trained and tested an artificial intelligence (AI) model for identifying substantial motion artifacts on CT pulmonary angiography (CTPA) that have a negative impact on diagnostic interpretation.MethodsWith IRB approval and HIPAA compliance, we queried our multicenter radiology report database (mPower, Nuance) for CTPA reports between July 2015 - March 2022 for the following terms: “motion artifacts,” “respiratory motion,” “technically inadequate,” and “suboptimal” or “limited exam.” All CTPA reports belonged to two quaternary (Site A, n= 335; B, n= 259) and a community (C, n= 199) healthcare sites. A thoracic radiologist reviewed CT images of all positive hits for motion artifacts (present or absent) and their severity (no diagnostic effect or major diagnostic impairment). Coronal multiplanar images belonging to 793 CTPA exams were de-identified and exported offline into an AI model building prototype (Cognex Vision Pro, Cognex Corporation) to train an AI model to perform two-class classification (“motion” or “no motion”) with data from the three sites (70% training dataset, n= 554; 30% validation dataset, n= 239). Separately, data from Site A and Site C were used for training and validating; testing was performed on the Site B CTPA exams. A 5-fold repeated cross-validation was performed to evaluate the model performance with accuracy and receiver operating characteristics analysis (ROC).ResultsAmong the CTPA images from 793 patients (mean age 63 ± 17 years; 391 males, 402 females), 372 had no motion artifacts, and 421 had substantial motion artifacts. The statistics for the average performance of the AI model after 5-fold repeated cross-validation for the two-class classification included 94% sensitivity, 91% specificity, 93% accuracy, and 0.93 area under the ROC curve (AUC: 95% CI 0.89-0.97).ConclusionThe AI model used in this study can successfully identify CTPA exams with diagnostic interpretation limiting motion artifacts in multicenter training and test datasets.Clinical relevanceThe AI model used in the study can help alert the technologists about the presence of substantial motion artifacts on CTPA where a repeat image acquisition can help salvage diagnostic information.

Non-contrast head CT (NCCT) is extremely insensitive for early (2 > 0.98).

We introduce Ego4D, a massive-scale egocentric video dataset and benchmark suite. It offers 3,670 hours of daily-life activity video spanning hundreds of scenarios (household, outdoor, workplace, leisure, etc.) captured by 931 unique camera wearers from 74 worldwide locations and 9 different countries. The approach to collection is designed to uphold rigorous privacy and ethics standards with consenting participants and robust de-identification procedures where relevant. Ego4D dramatically expands the volume of diverse egocentric video footage publicly available to the research community. Portions of the video are accompanied by audio, 3D meshes of the environment, eye gaze, stereo, and/or synchronized videos from multiple egocentric cameras at the same event. Furthermore, we present a host of new benchmark challenges centered around understanding the first-person visual experience in the past (querying an episodic memory), present (analyzing hand-object manipulation, audio-visual conversation, and social interactions), and future (forecasting activities). By publicly sharing this massive annotated dataset and benchmark suite, we aim to push the frontier of first-person perception. Project page: https://ego4d-data.org/, Comment: To appear in the Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022. This version updates the baseline result numbers for the Hands and Objects benchmark (appendix)

Background: Diffusion-weighted imaging (DWI) is the gold standard for detection of acute ischemic stroke (AIS) and may guide patient selection for endovascular therapy. Machine learning algorithms can enhance clinical workflow in suspected AIS by providing early diagnosis and quantitative imaging biomarkers. We tested the hypothesis that our deep learning algorithm for detection of acute ischemia performs robustly in a real-world environment. Methods: We developed a novel deep neural network architecture (YNet) that operates on MRI diffusion data at full resolution. YNet demonstrated strong performance in AIS detection and segmentation on test and validation datasets. After validation, we challenged YNet to assess 1081 consecutive brain MRIs acquired from 9 different scanners at our institution over a 2-week period. Studies were labeled as having AIS by a radiologist who used the final clinical imaging report. A series selection algorithm was applied to each study to identify DWI and apparent diffusion coefficient (ADC) series for input to the YNet using DICOM metadata to ensure relevant criteria are met (B value = 1000, axial plane, etc.). Patients with known diagnoses were excluded. Results: The series selection algorithm identified 826 studies that had DWI and ADC series with the established criteria. After exclusion of patients with known diagnoses, 378 studies were assessed by the YNet algorithm for the presence of AIS. The prevalence of AIS was 11.9% within this consecutive cohort. The algorithm detected 43 out of the 45 studies with AIS. Sensitivity and specificity for detection were 95.5% and 97.0%, respectively. Conclusion: Our YNet algorithm demonstrated excellent performance in detecting AIS in a serially-acquired dataset. Fully automated implementation will require further refinement of the series selection algorithm.

Background: Machine learning algorithms have proven accurate in the detection of intracranial hemorrhage (ICH) on head CT. Most reported algorithms, however, are limited to binary detection and global lesion volume estimation. We developed a pipeline to additionally perform subtype classification for clinical risk stratification, while also providing visual displays to improve interpretability of algorithm output (“black box problem”) and enable clinical systems integration. Methods: We developed a convolutional neural network (CNN) for ICH detection and subtype classification. The algorithm was trained on 2,209 head CT exams (hemorrhage-negative controls and the 5 major ICH subtypes) and outputs class-activation heatmaps to aid interpretability. We quantify intraparenchymal hemorrhage (IPH) volume using a modified U-Net CNN trained on a cohort of 110 manually-segmented hematomas to precisely delineate lesion boundaries. Algorithm output was stored as DICOM segmentation objects which can be overlaid on native images within our existing viewer applications, enabling the user to dynamically adjust characteristics such as color and transparency. Results: Diagnostic performance for ICH detection was high with ROC area-under-curve of 0.98. Fig 1A shows a heatmap highlighting different ICH foci. IPH subtype classification performance is summarized in Fig 1B with area-under-curve of 0.95. Fig 1C shows IPH segmentation and volume estimation overlays with automatically generated text for the medical record. Dice similarity coefficient for segmentation was 0.87. Conclusion: We describe a machine-learning approach for ICH detection, subtype classification, and IPH-specific volume estimation. Utilizing easily interpretable heatmaps, dynamically-adjustable overlays, and automated result text generation, our pipeline is capable of providing accurate information optimized for clinical integration.