Feasibility of a Curvature-based Enhanced Display System for Detecting Cerebral Aneurysms in MR Angiography

Geometric Deep Learning Using Vascular Surface Meshes for Modality-Independent Unruptured Intracranial Aneurysm Detection

Early detection of unruptured intracranial aneurysms (UIAs) enables better rupture risk and preventative treatment assessment. UIAs are usually diagnosed on Time-of-Flight Magnetic Resonance Angiographs (TOF-MRA) or contrast-enhanced Computed Tomography Angiographs (CTA). Various automatic voxel-based deep learning UIA detection methods have been developed, but these are limited to a single modality. We propose a modality-independent UIA detection method using a geometric deep learning model with high resolution surface meshes of brain vessels. A mesh convolutional neural network with ResU-Net style architecture was used. UIA detection performance was investigated with different input and pooling mesh resolutions, and including additional edge input features (shape index and curvedness). Both a higher resolution mesh (15,000 edges) and additional curvature edge features improved performance (average sensitivity: 65.6%, false positive count/image (FPC/image): 1.61). UIAs were detected in an independent TOF-MRA test set and a CTA test set with average sensitivity of 52.0% and 48.3% and average FPC/image of 1.04 and 1.05 respectively. We provide modality-independent UIA detection using a deep-learning vascular surface mesh model with comparable performance to state-of-the-art UIA detection methods.

Comparison of 1.5 T and 3 T magnetic resonance angiography for detecting cerebral aneurysms using deep learning-based computer-assisted detection software

PURPOSE

To compare the diagnostic performance of 1.5 T versus 3 T magnetic resonance angiography (MRA) for detecting cerebral aneurysms with clinically available deep learning-based computer-assisted detection software (EIRL aneurysm® [EIRL_an]), which has been approved by the Japanese Pharmaceuticals and Medical Devices Agency. We also sought to analyze the causes of potential false positives.

METHODS

In this single-center, retrospective study, we evaluated the MRA scans of 90 patients who underwent head MRA (1.5 T and 3 T in 45 patients each) in clinical practice. Overall, 51 patients had 70 aneurysms. We used MRI from a vendor not included in the dataset used to create the EIRL_an algorithm. Two radiologists determined the ground truth, the accuracy of the candidates noted by EIRL_an, and the causes of false positives. The sensitivity, number of false positives per case (FPs/case), and the causes of false positives were compared between 1.5 T and 3 T MRA. Pearson's χ2 test, Fisher's exact test, and the Mann‒Whitney U test were used for the statistical analyses as appropriate.

RESULTS

The sensitivity was high for 1.5 T and 3 T MRA (0.875‒1), but the number of FPs/case was significantly higher with 3 T MRA (1.511 vs. 2.578, p < 0.001). The most common causes of false positives (descending order) were the origin/bifurcation of vessels/branches, flow-related artifacts, and atherosclerosis and were similar between 1.5 T and 3 T MRA.

CONCLUSION

EIRL_an detected significantly more false-positive lesions with 3 T than with 1.5 T MRA in this external validation study. Our data may help physicians with limited experience with MRA to correctly diagnose aneurysms using EIRL_an.

Effectiveness of tuning an artificial intelligence algorithm for cerebral aneurysm diagnosis: a study of 10,000 consecutive cases

Diagnostic image analysis for unruptured cerebral aneurysms using artificial intelligence has a very high sensitivity. However, further improvement is needed because of a relatively high number of false positives. This study aimed to confirm the clinical utility of tuning an artificial intelligence algorithm for cerebral aneurysm diagnosis. We extracted 10,000 magnetic resonance imaging scans of participants who underwent brain screening using the "Brain Dock" system. The sensitivity and false positives/case for aneurysm detection were compared before and after tuning the algorithm. The initial diagnosis included only cases for which feedback to the algorithm was provided. In the primary analysis, the sensitivity of aneurysm diagnosis decreased from 96.5 to 90% and the false positives/case improved from 2.06 to 0.99 after tuning the algorithm (P < 0.001). In the secondary analysis, the sensitivity of aneurysm diagnosis decreased from 98.8 to 94.6% and the false positives/case improved from 1.99 to 1.03 after tuning the algorithm (P < 0.001). The false positives/case reduced without a significant decrease in sensitivity. Using large clinical datasets, we demonstrated that by tuning the algorithm, we could significantly reduce false positives with a minimal decline in sensitivity.

Detection of intracranial aneurysms using deep learning-based CAD system: usefulness of the scores of CNN's final layer for distinguishing between aneurysm and infundibular dilatation

PURPOSE

We evaluated the diagnostic performance of a clinically available deep learning-based computer-assisted diagnosis software for detecting unruptured aneurysms (UANs) using magnetic resonance angiography and assessed the functionality of the convolutional neural network (CNN) final layer score for distinguishing between UAN and infundibular dilatation (ID).

MATERIALS AND METHODS

EIRL brain aneurysm (EIRL_BA) was used in this study. The subjects were 117 UAN and/or ID cases including 100 UAN lesions (average sizes of 2.56 ± 1.45 mm) and 40 ID lesions (average sizes of 1.75 ± 0.41 mm) in any of internal carotid artery, middle cerebral artery, and anterior communicating artery, and 123 normal controls. The sensitivity, specificity, and accuracy of EIRL_BA were determined for UAN and ID or UAN only. Furthermore, the relationship between the lesion category and score was examined using a linear regression analysis model, and the receiver operating characteristic (ROC) analysis was used to assess whether the scores represent UAN-like characteristics.

RESULTS

EIRL_BA showed a total of 203 candidates (an average of 1.73/case) in UAN and/or ID cases and 98 candidates (an average of 0.80/case) in normal controls. For diagnosing either UAN/ID, EIRL_BA showed an overall sensitivity of 80%, specificity of 84.2%, and accuracy of 83.7%, resulting in the positive likelihood ratio of 5.0. For diagnosing UAN only, the overall sensitivity of 89.0, specificity of 82.6%, and accuracy of 83.2% resulting in the positive likelihood ratio of 5.1. In a linear regression analysis, the scores significantly increased in the candidates' first and second ranks in UAN (p < 0.05) but not in ID. An ROC analysis using the score for diagnosing UAN showed an area under the curve of 0.836.

CONCLUSION

EIRL_BA is applicable for detecting small UAN, and the CNN's final layer scores may be an effective index for discriminating UAN and ID and representing the likelihood of UAN.

Incidental cerebral aneurysms detected by a computer-assisted detection system based on artificial intelligence: A case series

RATIONALE

Computer-assisted detection (CAD) systems based on artificial intelligence (AI) using convolutional neural network (CNN) have been successfully used for the diagnosis of unruptured cerebral aneurysms in experimental situations. However, it is yet unclear whether CAD systems can detect cerebral aneurysms effectively in real-life clinical situations. This paper describes the diagnostic efficacy of CAD systems for cerebral aneurysms and the types of cerebral aneurysms that they can detect.

PATIENT CONCERNS

From March 7, 2017 to August 26, 2018 we performed brain magnetic resonance imaging (MRI) scans for 1623 subjects, to rule out intracranial diseases. We retrospectively reviewed the medical records including the history and images for each patient.

DIAGNOSES, INTERVENTIONS AND OUTCOMES

Among them, we encountered 5 cases in whom the cerebral aneurysms had been overlooked in the first and second round of imaging, and were detected for the first time by CAD. All missed aneurysms were less than 2 mm in diameter. Of the 5 aneurysms, 2 were internal carotid artery (ICA) paraclinoid aneurysms, 2 were Internal carotid-posterior communicating artery (IC-PC) aneurysms and 1 was a distal middle cerebral artery (MCA) aneurysm.

LESSONS

Our CAD system can detect very small aneurysms masked by the surrounding arteries and difficult for radiologists to detect. In the future, CAD systems might pave the way to substitute the workload of diagnostic radiologists and reduce the cost of human labor.

Deep Learning for MR Angiography: Automated Detection of Cerebral Aneurysms

Purpose To develop and evaluate a supportive algorithm using deep learning for detecting cerebral aneurysms at time-of-flight MR angiography to provide a second assessment of images already interpreted by radiologists. Materials and Methods MR images reported by radiologists to contain aneurysms were extracted from four institutions for the period from November 2006 through October 2017. The images were divided into three data sets: training data set, internal test data set, and external test data set. The algorithm was constructed by deep learning with the training data set, and its sensitivity to detect aneurysms in the test data sets was evaluated. To find aneurysms that had been overlooked in the initial reports, two radiologists independently performed a blinded interpretation of aneurysm candidates detected by the algorithm. When there was disagreement, the final diagnosis was made in consensus. The number of newly detected aneurysms was also evaluated. Results The training data set, which provided training and validation data, included 748 aneurysms (mean size, 3.1 mm ± 2.0 [standard deviation]) from 683 examinations; 318 of these examinations were on male patients (mean age, 63 years ± 13) and 365 were on female patients (mean age, 64 years ± 13). Test data were provided by the internal test data set (649 aneurysms [mean size, 4.1 mm ± 3.2] in 521 examinations, including 177 male patients and 344 female patients with mean age of 66 years ± 12 and 67 years ± 13, respectively) and the external test data set (80 aneurysms [mean size, 4.1 mm ± 2.1] in 67 examinations, including 19 male patients and 48 female patients with mean age of 63 years ± 12 and 68 years ± 12, respectively). The sensitivity was 91% (592 of 649) and 93% (74 of 80) for the internal and external test data sets, respectively. The algorithm improved aneurysm detection in the internal and external test data sets by 4.8% (31 of 649) and 13% (10 of 80), respectively, compared with the initial reports. Conclusion A deep learning algorithm detected cerebral aneurysms in radiologic reports with high sensitivity and improved aneurysm detection compared with the initial reports. © RSNA, 2018 See also the editorial by Flanders in this issue.

Computer-Assisted Detection of Cerebral Aneurysms in MR Angiography in a Routine Image-Reading Environment: Effects on Diagnosis by Radiologists

BACKGROUND AND PURPOSE

Experiences with computer-assisted detection of cerebral aneurysms in diagnosis by radiologists in real-life clinical environments have not been reported. The purpose of this study was to evaluate the usefulness of computer-assisted detection in a routine reading environment.

MATERIALS AND METHODS

During 39 months in a routine clinical practice environment, 2701 MR angiograms were each read by 2 radiologists by using a computer-assisted detection system. Initial interpretation was independently made without using the detection system, followed by a possible alteration of diagnosis after referring to the lesion candidate output from the system. We used the final consensus of the 2 radiologists as the reference standard. The sensitivity and specificity of radiologists before and after seeing the lesion candidates were evaluated by aneurysm- and patient-based analyses.

RESULTS

The use of the computer-assisted detection system increased the number of detected aneurysms by 9.3% (from 258 to 282). Aneurysm-based analysis revealed that the apparent sensitivity of the radiologists' diagnoses made without and with the detection system was 64% and 69%, respectively. The detection system presented 82% of the aneurysms. The detection system more frequently benefited radiologists than being detrimental.

CONCLUSIONS

Routine integration of computer-assisted detection with MR angiography for cerebral aneurysms is feasible, and radiologists can detect a number of additional cerebral aneurysms by using the detection system without a substantial decrease in their specificity. The low confidence of radiologists in the system may limit its usefulness.

Computer-aided diagnosis for detection of lacunar infarcts on MR images: ROC analysis of radiologists' performance

The purpose of this study was to retrospectively evaluate radiologist performance in detection of lacunar infarcts on T1- and T2-weighted images, without and with the use of a computer-aided diagnosis (CAD) scheme. Thirty T1-weighted and 30 T2-weighted MR images obtained from 30 patients were used for assessing observer performance. These images were acquired using the fast spin-echo sequence with a 1.5-T MR imaging scanner. The group included 15 patients (age range, 48-83 years; mean age, 67.2 years; 10 men and five women) with a lacunar infarct and 15 patients (age range, 39-76 years; mean age, 64.0 years; eight men and seven women) without lacunar infarcts. Nine radiologists participated in the study. The radiologists initially interpreted the T1- and T2-weighted images without and then with the use of CAD, which indicated their confidence levels regarding the presence (or absence) of lacunar infarcts and the most likely position of a lesion on each MR scan. The observers' performance without and with the computer output was evaluated by performing receiver operating characteristic analysis. For the nine radiologists, the mean area under the best-fit binormal receiver operating characteristic curve plotted for unit square values of radiologists who interpreted the images without and with the scheme were 0.891 and 0.937, respectively. The performance of the radiologists improved significantly when they used the computer output (p=0.032). The CAD scheme has potential to improve the accuracy of radiologists' performance in detection of lacunar infarcts.

Magnetic Resonance Image Analysis for Brain CAD Systems with Machine Learning

This chapter describes the image analysis for brain Computer-Aided Diagnosis (CAD) systems with machine learning techniques, which could assist radiologists in the detection of such brain diseases as asymptomatic unruptured aneurysms, Alzheimer’s Disease (AD), vascular dementia, and Multiple Sclerosis (MS) by magnetic resonance imaging. Image analysis in CAD systems consists of image enhancement, initial detection, and image feature extraction, including segmentation. In addition, the authors review the classification of true and false positives using machine learning techniques, as well as the evaluation methods and development cycle for CAD systems.

Computer-aided diagnosis scheme for detection of lacunar infarcts on MR images

RATIONALE AND OBJECTIVES

The detection and management of asymptomatic lacunar infarcts on magnetic resonance (MR) images are important tasks for radiologists to ensure the prevention of severe cerebral infarctions. However, accurate identification of the lacunar infarcts on MR images is a difficult task for the radiologists. Therefore the purpose of this study was to develop a computer-aided diagnosis scheme for the detection of lacunar infarcts to assist radiologists' interpretation as a "second opinion."

MATERIALS AND METHODS

Our database comprised 1,143 T1- and 1,143 T2-weighted images obtained from 132 patients. The locations of the lacunar infarcts were determined by experienced neuroradiologists. We first segmented the cerebral region in a T1-weighted image by using a region growing technique for restricting the search area of lacunar infarcts. For identifying the initial lacunar infarcts candidates, a top-hat transform and multiple-phase binarization were then applied to the T2-weighted image within the segmented cerebral region. For eliminating the false positives (FPs), we determined 12 features--the locations x and y, signal intensity differences in the T1- and T2-weighted images, nodular components from a scale of 1 to 4, and nodular and linear components from a scale of 1 to 4. The nodular components and the linear components were obtained using a filter bank technique. The rule-based schemes and a support vector machine with 12 features were applied to the regions of the initial candidates for distinguishing between lacunar infarcts and FPs.

RESULTS

Our computerized scheme was evaluated by using a holdout method. The sensitivity of the detection of lacunar infarcts was 96.8% (90/93) with 0.76 FP per image.

CONCLUSIONS

Our computerized scheme would be useful in assisting radiologists for identifying lacunar infarcts in MR images.