In Journal of stroke and cerebrovascular diseases : the official journal of National Stroke Association

OBJECTIVES : Moyamoya disease is a unique cerebrovascular disorder that is characterized by chronic bilateral stenosis of the internal carotid arteries and by the formation of an abnormal vascular network called moyamoya vessels. In this stury, the authors inspected whether differentiation between patients with moyamoya disease and those with atherosclerotic disease or normal controls might be possible by using deep machine learning technology.

MATERIALS AND METHODS : This study included 84 consecutive patients diagnosed with moyamoya disease at our hospital between April 2009 and July 2016. In each patient, two axial continuous slices of T2-weighed imaging at the level of the basal cistern, basal ganglia, and centrum semiovale were acquired. The image sets were processed by using code written in the programming language Python 3.7. Deep learning with fine tuning developed using VGG16 comprised several layers.

RESULTS : The accuracies of distinguishing between patients with moyamoya disease and those with atherosclerotic disease or controls in the basal cistern, basal ganglia, and centrum semiovale levels were 92.8, 84.8, and 87.8%, respectively.

CONCLUSION : The authors showed excellent results in terms of accuracy of differential diagnosis of moyamoya disease using AI with the conventional T2 weighted images. The authors suggest the possibility of diagnosing moyamoya disease using AI technique and demonstrate the area of interest on which AI focuses while processing magnetic resonance images.

Akiyama Yukinori, Mikami Takeshi, Mikuni Nobuhiro

2020-Sep-25

Artificial intelligence, Deep learning, Diagnostic accuracy, Moyamoya disease

In Journal of stroke and cerebrovascular diseases : the official journal of National Stroke Association

BACKGROUND AND PURPOSE : Accurate prediction using simple and changeable variables is clinically meaningful because some known-predictors, such as stroke severity and patients age cannot be modified with rehabilitative treatment. There are limited clinical prediction rules (CPRs) that have been established using only changeable variables to predict the activities of daily living (ADL) dependence of stroke patients. This study aimed to develop and assess the CPRs using machine learning-based methods to identify ADL dependence in stroke patients.

METHODS : In total, 1125 stroke patients were investigated. We used a maintained database of all stroke patients who were admitted to the convalescence rehabilitation ward of our facility. The classification and regression tree (CART) methodology with only the FIM subscores was used to predict the ADL dependence.

RESULTS : The CART method identified FIM transfer (bed, chair, and wheelchair) (score ≤ 4.0 or > 4.0) as the best single discriminator for ADL dependence. Among those with FIM transfer (bed, chair, and wheelchair) score > 4.0, the next best predictor was FIM bathing (score ≤ 2.0 or > 2.0). Among those with FIM transfer (bed, chair, and wheelchair) score ≤ 4.0, the next predictor was FIM transfer toilet (score ≤ 3 or > 3). The accuracy of the CART model was 0.830 (95% confidence interval, 0.804-0.856).

CONCLUSION : Machine learning-based CPRs with moderate predictive ability for the identification of ADL dependence in the stroke patients were developed.

Iwamoto Yuji, Imura Takeshi, Tanaka Ryo, Imada Naoki, Inagawa Tetsuji, Araki Hayato, Araki Osamu

2020-Sep-26

Activities of daily living, Decision-tree analysis, Prediction, Rehabilitation, Stroke

In Journal of stroke and cerebrovascular diseases : the official journal of National Stroke Association

OBJECTIVE : To build radiomic model in differentiating dissecting aneurysm (DA) from complicated saccular aneurysm (SA) based on high-resolution magnetic resonance imaging (HR-MRI) through machine-learning algorithm.

METHODS : Overall, 851 radiomic features from 77 cases were retrospectively analyzed, and the ElasticNet algorithm was used to build the radiomic model. A clinico-radiological model using clinical features and conventional MRI findings was also built. An integrated model was then built by incorporating the radiomic model and clinico-radiological model. The diagnostic abilities of these models were evaluated using leave one out cross validation and quantified using the receiver operating characteristic (ROC) analysis. The diagnostic performance of radiologists was also evaluated for comparison.

RESULTS : Five features were used to form the radiomic model, which yielded an area under the ROC curve (AUC) of 0.912 (95 % CI 0.846-0.976), sensitivity of 0.852, and specificity of 0.861. The radiomic model achieved a better diagnostic performance than the clinico-radiological model (AUC=0.743, 95 % CI 0.623-0.862), integrated model (AUC=0.888, 95 % CI 0.811-0.965), and even many radiologists.

CONCLUSION : Radiomic features derived from HR-MRI can reliably be used to build a radiomic model for effectively differentiating between DA and complicated SA, and it can provide an objective basis for the selection of clinical treatment plan.

Cao Xin, Xia Wei, Tang Ye, Zhang Bo, Yang Jinming, Zeng Yanwei, Geng Daoying, Zhang Jun

2020-Sep-08

Aneurysm, High-resolution magnetic resonance imaging, Machine-learning, Radiomics

In Computers in biology and medicine

BACKGROUND : Hypertension (HPT) occurs when there is increase in blood pressure (BP) within the arteries, causing the heart to pump harder against a higher afterload to deliver oxygenated blood to other parts of the body.

PURPOSE : Due to fluctuation in BP, 24-h ambulatory blood pressure monitoring has emerged as a useful tool for diagnosing HPT but is limited by its inconvenience. So, an automatic diagnostic tool using electrocardiogram (ECG) signals is used in this study to detect HPT automatically.

METHOD : The pre-processed signals are fed to a convolutional neural network model. The model learns and identifies unique ECG signatures for classification of normal and hypertension ECG signals. The proposed model is evaluated by the 10-fold and leave one out patient based validation techniques.

RESULTS : A high classification accuracy of 99.99% is achieved for both validation techniques. This is one of the first few studies to have employed deep learning algorithm coupled with ECG signals for the detection of HPT. Our results imply that the developed tool is useful in a hospital setting as an automated diagnostic tool, enabling the effortless detection of HPT using ECG signals.

Soh Desmond Chuang Kiat, Ng E Y K, Jahmunah V, Oh Shu Lih, Tan Ru San, Acharya U Rajendra

2020-Sep-17

10-Fold validation, Automated diagnostic tool, Convolutional neural network, Hypertension, Leave one patient out validation, Masked hypertension

In International journal of medical informatics ; h5-index 49.0

OBJECTIVE : This study aims to develop and test a new computer-aided diagnosis (CAD) scheme of chest X-ray images to detect coronavirus (COVID-19) infected pneumonia.

METHOD : CAD scheme first applies two image preprocessing steps to remove the majority of diaphragm regions, process the original image using a histogram equalization algorithm, and a bilateral low-pass filter. Then, the original image and two filtered images are used to form a pseudo color image. This image is fed into three input channels of a transfer learning-based convolutional neural network (CNN) model to classify chest X-ray images into 3 classes of COVID-19 infected pneumonia, other community-acquired no-COVID-19 infected pneumonia, and normal (non-pneumonia) cases. To build and test the CNN model, a publicly available dataset involving 8474 chest X-ray images is used, which includes 415, 5179 and 2,880 cases in three classes, respectively. Dataset is randomly divided into 3 subsets namely, training, validation, and testing with respect to the same frequency of cases in each class to train and test the CNN model.

RESULTS : The CNN-based CAD scheme yields an overall accuracy of 94.5 % (2404/2544) with a 95 % confidence interval of [0.93,0.96] in classifying 3 classes. CAD also yields 98.4 % sensitivity (124/126) and 98.0 % specificity (2371/2418) in classifying cases with and without COVID-19 infection. However, without using two preprocessing steps, CAD yields a lower classification accuracy of 88.0 % (2239/2544).

CONCLUSION : This study demonstrates that adding two image preprocessing steps and generating a pseudo color image plays an important role in developing a deep learning CAD scheme of chest X-ray images to improve accuracy in detecting COVID-19 infected pneumonia.

Heidari Morteza, Mirniaharikandehei Seyedehnafiseh, Khuzani Abolfazl Zargari, Danala Gopichandh, Qiu Yuchen, Zheng Bin

2020-Sep-23

COVID-19 diagnosis, Computer-aided diagnosis, Convolution neural network (CNN), Coronavirus, Disease classification, VGG16 network

In Mathematical biosciences and engineering : MBE

Everett Rebecca, Flores Kevin B, Henscheid Nick, Lagergren John, Larripa Kamila, Li Ding, Nardini John T, Nguyen Phuong T T, Pitman E Bruce, Rutter Erica M

2020-May-19

** Bayesian estimation , cancer heterogeneity , generative adversarial networks , glioblastoma multiforme , machine learning , mathematical oncology , non-parametric estimation , nonlinear mixed effects , spatiotemporal data , tumor growth , variational autoencoders , virtual populations **