Expert-level detection of acute intracranial hemorrhage on head computed tomography using deep learning

Significance Computed tomography (CT) of the head is the workhorse medical imaging modality used worldwide to diagnose neurologic emergencies. However, these gray scale images are limited by low signal-to-noise, poor contrast, and a high incidence of image artifacts. A unique challenge is to identify tiny subtle abnormalities in a large 3D volume with near-perfect sensitivity. We used a single-stage, end-to-end, fully convolutional neural network to achieve accuracy levels comparable to that of highly trained radiologists, including both identification and localization of abnormalities that are missed by radiologists. Computed tomography (CT) of the head is used worldwide to diagnose neurologic emergencies. However, expertise is required to interpret these scans, and even highly trained experts may miss subtle life-threatening findings. For head CT, a unique challenge is to identify, with perfect or near-perfect sensitivity and very high specificity, often small subtle abnormalities on a multislice cross-sectional (three-dimensional [3D]) imaging modality that is characterized by poor soft tissue contrast, low signal-to-noise using current low radiation-dose protocols, and a high incidence of artifacts. We trained a fully convolutional neural network with 4,396 head CT scans performed at the University of California at San Francisco and affiliated hospitals and compared the algorithm’s performance to that of 4 American Board of Radiology (ABR) certified radiologists on an independent test set of 200 randomly selected head CT scans. Our algorithm demonstrated the highest accuracy to date for this clinical application, with a receiver operating characteristic (ROC) area under the curve (AUC) of 0.991 ± 0.006 for identification of examinations positive for acute intracranial hemorrhage, and also exceeded the performance of 2 of 4 radiologists. We demonstrate an end-to-end network that performs joint classification and segmentation with examination-level classification comparable to experts, in addition to robust localization of abnormalities, including some that are missed by radiologists, both of which are critically important elements for this application.

[1]  Mark Craven,et al.  Active Learning with Real Annotation Costs , 2008 .

[2]  Hao Chen,et al.  Gland segmentation in colon histology images: The glas challenge contest , 2016, Medical Image Anal..

[3]  Yogesh Rathi,et al.  Harmonizing Diffusion MRI Data Across Magnetic Field Strengths , 2018, MICCAI.

[4]  Katrin Tomanek,et al.  Resource-aware annotation through active learning , 2010 .

[5]  Likang Xu,et al.  Traumatic Brain Injury–Related Emergency Department Visits, Hospitalizations, and Deaths — United States, 2007 and 2013 , 2017, Morbidity and mortality weekly report. Surveillance summaries.

[6]  Albert C. S. Chung,et al.  Deep supervision with additional labels for retinal vessel segmentation task , 2018, MICCAI.

[7]  Thomas A. Funkhouser,et al.  Dilated Residual Networks , 2017, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[8]  Brandon K. Fornwalt,et al.  Advanced machine learning in action: identification of intracranial hemorrhage on computed tomography scans of the head with clinical workflow integration , 2018, npj Digital Medicine.

[9]  Hongzhi Wang,et al.  A Multi-atlas Approach to Region of Interest Detection for Medical Image Classification , 2017, MICCAI.

[10]  Luc Van Gool,et al.  The Pascal Visual Object Classes (VOC) Challenge , 2010, International Journal of Computer Vision.

[11]  Ewout W Steyerberg,et al.  Prediction of Outcome in Traumatic Brain Injury with Computed Tomographic Characteristics: A Comparison between the Computed Tomographic Classification and Combinations of Computed Tomographic Predictors , 2005, Neurosurgery.

[12]  Fisher Cm,et al.  Relation of cerebral vasospasm to subarachnoid hemorrhage visualized by computerized tomographic scanning. , 1980, Neurosurgery.

[13]  William A. Gale,et al.  A sequential algorithm for training text classifiers , 1994, SIGIR '94.

[14]  Jitendra Malik,et al.  Cost-Sensitive Active Learning for Intracranial Hemorrhage Detection , 2018, MICCAI.

[15]  Richard D. White,et al.  Automated Critical Test Findings Identification and Online Notification System Using Artificial Intelligence in Imaging. , 2017, Radiology.

[16]  Andrew H. Beck,et al.  Diagnostic Assessment of Deep Learning Algorithms for Detection of Lymph Node Metastases in Women With Breast Cancer , 2017, JAMA.

[17]  Subhashini Venugopalan,et al.  Development and Validation of a Deep Learning Algorithm for Detection of Diabetic Retinopathy in Retinal Fundus Photographs. , 2016, JAMA.

[18]  Chao Lu,et al.  Retrospective study , 2016, Medicine.

[19]  David K. Menon,et al.  Changing patterns in the epidemiology of traumatic brain injury , 2013, Nature Reviews Neurology.

[20]  Geraint Rees,et al.  Clinically applicable deep learning for diagnosis and referral in retinal disease , 2018, Nature Medicine.

[21]  J. Claude Hemphill,et al.  Management of intracerebral hemorrhage. , 2017, Handbook of clinical neurology.

[22]  Hengyong Yu,et al.  Convolutional Neural Network Based Metal Artifact Reduction in X-Ray Computed Tomography , 2017, IEEE Transactions on Medical Imaging.

[23]  Kaiming He,et al.  Mask R-CNN , 2017, 2017 IEEE International Conference on Computer Vision (ICCV).

[24]  Trevor Darrell,et al.  Fully Convolutional Networks for Semantic Segmentation , 2017, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[25]  David Cohn,et al.  Active Learning , 2010, Encyclopedia of Machine Learning.

[26]  Sasank Chilamkurthy,et al.  Deep learning algorithms for detection of critical findings in head CT scans: a retrospective study , 2018, The Lancet.

[27]  Ali Abdalvand,et al.  Interpretation of Computed Tomography of the Head: Emergency Physicians versus Radiologists , 2013, Trauma monthly.

[28]  H. Sebastian Seung,et al.  Query by committee , 1992, COLT '92.

[29]  Jamshid Ghajar,et al.  Surgical management of acute subdural hematomas. , 2006, Neurosurgery.

[30]  Silvio Savarese,et al.  Active Learning for Convolutional Neural Networks: A Core-Set Approach , 2017, ICLR.

[31]  Anthony Marmarou,et al.  A new classification of head injury based on computerized tomography , 1991 .

[32]  T Brott,et al.  The ABCs of measuring intracerebral hemorrhage volumes. , 1996, Stroke.

[33]  Jitendra Malik,et al.  PatchFCN for Intracranial Hemorrhage Detection , 2018, ArXiv.

[34]  A. J. M. van der Werf,et al.  Surgical management of acute subdural hematomas , 1975, Clinical Neurology and Neurosurgery.

[35]  L. Sugrue,et al.  Hybrid 3D/2D Convolutional Neural Network for Hemorrhage Evaluation on Head CT , 2018, American Journal of Neuroradiology.

[36]  N J Pelc,et al.  Nonlinear partial volume artifacts in x-ray computed tomography. , 1980, Medical physics.

[37]  Mohammad Mansouri,et al.  An explainable deep-learning algorithm for the detection of acute intracranial haemorrhage from small datasets , 2018, Nature Biomedical Engineering.

[38]  Lin Yang,et al.  Coarse-to-Fine Stacked Fully Convolutional Nets for lymph node segmentation in ultrasound images , 2016, 2016 IEEE International Conference on Bioinformatics and Biomedicine (BIBM).

[39]  Konstantinos Kamnitsas,et al.  Autofocus Layer for Semantic Segmentation , 2018, MICCAI.

[40]  Lin Yang,et al.  Suggestive Annotation: A Deep Active Learning Framework for Biomedical Image Segmentation , 2017, MICCAI.

[41]  J. Weinberger,et al.  Management of intracerebral hemorrhage , 2007, Vascular health and risk management.

[42]  T. Kurki,et al.  Reliability of diagnosis of traumatic brain injury by computed tomography in the acute phase. , 2009, Journal of neurotrauma.

[43]  Sebastian Thrun,et al.  Dermatologist-level classification of skin cancer with deep neural networks , 2017, Nature.

[44]  David S. Rosen,et al.  Subarachnoid hemorrhage grading scales , 2005, Neurocritical care.

[45]  E. Mohammadi,et al.  Barriers and facilitators related to the implementation of a physiological track and trigger system: A systematic review of the qualitative evidence , 2017, International journal for quality in health care : journal of the International Society for Quality in Health Care.

[46]  Marcus A. Badgeley,et al.  Automated deep-neural-network surveillance of cranial images for acute neurologic events , 2018, Nature Medicine.

[47]  Jian Sun,et al.  Object Detection Networks on Convolutional Feature Maps , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[48]  D. Ballard,et al.  False-negative Interpretations of Cranial Computed Tomography in Aneurysmal Subarachnoid Hemorrhage. , 2016, Academic emergency medicine : official journal of the Society for Academic Emergency Medicine.