AI prediction of extracorporeal shock wave lithotripsy outcomes for ureteral stones by machine learning-based analysis with a variety of stone and patient characteristics.

We propose an artificial intelligence prediction method for extracorporeal shock wave lithotripsy treatment outcomes by analysis of a wide variety of variables. We retrospectively reviewed the records of 171 patients from between January 2009 and November 2019 that underwent shock wave lithotripsy at Wakayama Medical University, Japan, for ureteral stones shown on preoperative non-contrast computed tomography. This prediction method consisted of stone area extraction, stone analyzing factor extraction from non-contrast computed tomography images, and shock wave lithotripsy treatment result prediction by a non-linear support vector machine for analysis of 15 input and automatic measurement factors. Input factors included patient age, skin-to-stone distance, and maximum ureteral wall thickness, and the automatic measurement factors included 11 non-contrast computed tomography image texture factors in the stone area and stone volume. Permutation feature importance was also applied to the artificial intelligence prediction results to analyze the importance of each factor relating to estimate decision grounds. The prediction performance was evaluated by five-fold cross-validation, it obtained 0.742 of the mean area under the receiver operating characteristic curve. The proposed method is shown by these results to have robust data diversity and effective clinical application. As a result of permutation feature importance, some factors that showed high p-values in the significant difference tests were thought to have a high contribution to the proposed prediction method. Future issues include validation using a larger volume of high-resolution clinical non-contrast computed tomography image data and the application of deep learning.

Automated volume measurement of abdominal adipose tissue from entire abdominal cavity in Dixon MR images using deep learning.

The purpose of this study was to realize an automated volume measurement of abdominal adipose tissue from the entire abdominal cavity in Dixon magnetic resonance (MR) images using deep learning. Our algorithm involves a combination of extraction of the abdominal cavity and body trunk regions using deep learning and extraction of a fat region based on automatic thresholding. To evaluate the proposed method, we calculated the Dice coefficient (DC) between the extracted regions using deep learning and labeled images. We also compared the visceral adipose tissue (VAT) and subcutaneous adipose tissue volumes calculated by employing the proposed method with those calculated from computed tomography (CT) images scanned on the same day using the automatic calculation method previously developed by our group. We implemented our method as a plug-in in a web-based medical image processing platform. The DCs of the abdominal cavity and body trunk regions were 0.952 ± 0.014 and 0.995 ± 0.002, respectively. The VAT volume measured from MR images using the proposed method was almost equivalent to that measured from CT images. The time required for our plug-in to process the test set was 118.9 ± 28.0 s. Using our proposed method, the VAT volume measured from MR images can be an alternative to that measured from CT images.

Automatic detection of primary and metastatic lesions on cervicothoracic region and whole-body bone using a uniform machine-learnable approach for [18F]-FDG-PET/CT image analysis.

We propose a method to detect primary and metastatic lesions with Fluorine-18 fluorodeoxyglucose (FDG) accumulation in the lung field, neck, mediastinum, and bony regions on the FDG-PET/CT images. To search for systemic lesions, various anatomical structures must be considered. The proposed method is addressed by using an extraction process for anatomical regions and a uniform lesion detection approach. The uniform approach does not utilize processes that reflect any region-specific anatomical aspects but has a machine-learnable framework. Therefore, it can work as a lesion detection process for a specific anatomical region if it machine-learns the specific region data. In this study, three lesion detection processes for the whole-body bone region, lung field, or neck-mediastinum region are obtained. These detection processes include lesion candidate detection and false positive (FP) candidate elimination. The lesion candidate detection is based on a voxel anomaly detection with a one-class support vector machine. The FP candidate elimination is performed using an AdaBoost classifier ensemble. The image features used by the ensemble are selected sequentially during training and are optimal for candidate classification. Three-fold cross-validation was used to detect performance with the 54 diseased FDG-PET/CT images. The mean sensitivity for detecting primary and metastatic lesions at 3 FPs per case was 0.89 with a 0.10 standard deviation (SD) in the bone region, 0.80 with a 0.10 SD in the lung field, and 0.87 with a 0.10 SD in the neck region. The average areas under the ROC curve were 0.887 with a 0.125 SD for detecting bone metastases, 0.900 with a 0.063 SD for detecting pulmonary lesions, and 0.927 with a 0.035 SD for detecting the neck-mediastinum lesions. These detection performances indicate that the proposed method could be applied clinically. These results also show that the uniform approach has high versatility for providing various lesion detection processes.

Utility of U-Net for the objective segmentation of the fibroglandular tissue region on clinical digital mammograms.

This study investigates the equivalence or compatibility between U-Net and visual segmentations of fibroglandular tissue regions by mammography experts for calculating the breast density and mean glandular dose (MGD). A total of 703 mediolateral oblique-view mammograms were used for segmentation. Two region types were set as the ground truth (determined visually): (1) one type included only the region where fibroglandular tissue was identifiable (called the 'dense region'); (2) the other type included the region where the fibroglandular tissue may have existed in the past, provided that apparent adipose-only parts, such as the retromammary space, are excluded (the 'diffuse region'). U-Net was trained to segment the fibroglandular tissue region with an adaptive moment estimation optimiser, five-fold cross-validated with 400 training and 100 validation mammograms, and tested with 203 mammograms. The breast density and MGD were calculated using the van Engeland and Dance formulas, respectively, and compared between U-Net and the ground truth with the Dice similarity coefficient and Bland-Altman analysis. Dice similarity coefficients between U-Net and the ground truth were 0.895 and 0.939 for the dense and diffuse regions, respectively. In the Bland-Altman analysis, no proportional or fixed errors were discovered in either the dense or diffuse region for breast density, whereas a slight proportional error was discovered in both regions for the MGD (the slopes of the regression lines were -0.0299 and -0.0443 for the dense and diffuse regions, respectively). Consequently, the U-Net and ground truth were deemed equivalent (interchangeable) for breast density and compatible (interchangeable following four simple arithmetic operations) for MGD. U-Net-based segmentation of the fibroglandular tissue region was satisfactory for both regions, providing reliable segmentation for breast density and MGD calculations. U-Net will be useful in developing a reliable individualised screening-mammography programme, instead of relying on the visual judgement of mammography experts.

A review on AI in PET imaging.

Artificial intelligence (AI) has been applied to various medical imaging tasks, such as computer-aided diagnosis. Specifically, deep learning techniques such as convolutional neural network (CNN) and generative adversarial network (GAN) have been extensively used for medical image generation. Image generation with deep learning has been investigated in studies using positron emission tomography (PET). This article reviews studies that applied deep learning techniques for image generation on PET. We categorized the studies for PET image generation with deep learning into three themes as follows: (1) recovering full PET data from noisy data by denoising with deep learning, (2) PET image reconstruction and attenuation correction with deep learning and (3) PET image translation and synthesis with deep learning. We introduce recent studies based on these three categories. Finally, we mention the limitations of applying deep learning techniques to PET image generation and future prospects for PET image generation.

Machine Learning-Based Diagnosis in Laser Resonance Frequency Analysis for Implant Stability of Orthopedic Pedicle Screws.

Evaluation of the initial stability of implants is essential to reduce the number of implant failures of pedicle screws after orthopedic surgeries. Laser resonance frequency analysis (L-RFA) has been recently proposed as a viable diagnostic scheme in this regard. In a previous study, L-RFA was used to demonstrate the diagnosis of implant stability of monoaxial screws with a fixed head. However, polyaxial screws with movable heads are also frequently used in practice. In this paper, we clarify the characteristics of the laser-induced vibrational spectra of polyaxial screws which are required for making L-RFA diagnoses of implant stability. In addition, a novel analysis scheme of a vibrational spectrum using L-RFA based on machine learning is demonstrated and proposed. The proposed machine learning-based diagnosis method demonstrates a highly accurate prediction of implant stability (peak torque) for polyaxial pedicle screws. This achievement will contribute an important analytical method for implant stability diagnosis using L-RFA for implants with moving parts and shapes used in various clinical situations.

Automatic diagnosis of melanoma using hyperspectral data and GoogLeNet.

BACKGROUND: Melanoma is a type of superficial tumor. As advanced melanoma has a poor prognosis, early detection and therapy are essential to reduce melanoma-related deaths. To that end, there is a need to develop a quantitative method for diagnosing melanoma. This paper reports the development of such a diagnostic system using hyperspectral data (HSD) and a convolutional neural network, which is a type of machine learning. MATERIALS AND METHODS: HSD were acquired using a hyperspectral imager, which is a type of spectrometer that can simultaneously capture information about wavelength and position. GoogLeNet pre-trained with Imagenet was used to model the convolutional neural network. As many CNNs (including GoogLeNet) have three input channels, the HSD (involving 84 channels) could not be input directly. For that reason, a "Mini Network" layer was added to reduce the number of channels from 84 to 3 just before the GoogLeNet input layer. In total, 619 lesions (including 278 melanoma lesions and 341 non-melanoma lesions) were used for training and evaluation of the network. RESULTS AND CONCLUSION: The system was evaluated by 5-fold cross-validation, and the results indicate sensitivity, specificity, and accuracy of 69.1%, 75.7%, and 72.7% without data augmentation, 72.3%, 81.2%, and 77.2% with data augmentation, respectively. In future work, it is intended to improve the Mini Network and to increase the number of lesions.

AI approach of cycle-consistent generative adversarial networks to synthesize PET images to train computer-aided diagnosis algorithm for dementia.

OBJECTIVE: An artificial intelligence (AI)-based algorithm typically requires a considerable amount of training data; however, few training images are available for dementia with Lewy bodies and frontotemporal lobar degeneration. Therefore, this study aims to present the potential of cycle-consistent generative adversarial networks (CycleGAN) to obtain enough number of training images for AI-based computer-aided diagnosis (CAD) algorithms for diagnosing dementia. METHODS: We trained CycleGAN using 43 amyloid-negative and 45 positive images in slice-by-slice. RESULTS: The CycleGAN can be used to synthesize reasonable amyloid-positive images, and the continuity of slices was preserved. DISCUSSION: Our results show that CycleGAN has the potential to generate a sufficient number of training images for CAD of dementia.

Pilot study of eruption forecasting with muography using convolutional neural network.

Muography is a novel method of visualizing the internal structures of active volcanoes by using high-energy near-horizontally arriving cosmic muons. The purpose of this study is to show the feasibility of muography to forecast the eruption event with the aid of the convolutional neural network (CNN). In this study, seven daily consecutive muographic images were fed into the CNN to compute the probability of eruptions on the eighth day, and our CNN model was trained by hyperparameter tuning with the Bayesian optimization algorithm. By using the data acquired in Sakurajima volcano, Japan, as an example, the forecasting performance achieved a value of 0.726 for the area under the receiver operating characteristic curve, showing the reasonable correlation between the muographic images and eruption events. Our result suggests that muography has the potential for eruption forecasting of volcanoes.

Automatic delineation algorithm of reference region for amyloid imaging based on kinetics.

OBJECTIVE: This study aims to develop an algorithm named AutoRef to delineate a reference region for quantitative PET amyloid imaging. METHODS: AutoRef sets the reference region automatically using a distinguishing feature in the kinetics of reference region. This is reflected in the shapes of the tissue time activity curve. A statistical shape recognition algorithm of the gaussian mixture model is applied with considering spatial and temporal information on a reference region. We evaluate the BPND with manually set reference region and AutoRef using 86 cases (43 positive cases, 10 equivocal cases, and 33 negative cases) of dynamically scanned 11C-Pittsburgh Compound-B. RESULTS: From the Bland-Altman plot, the difference between two BPND is 0.099 ± 0.21 as standard deviation, and no significant systematic error is observed between the BPND with AutoRef and with manual definition of a reference region. Although a proportional error is detected, it is smaller than the 95% limits of agreement. Therefore, the proportional error is negligibly small. CONCLUSIONS: AutoRef presents the same performance as the manual definition of the reference region. Further, since AutoRef is more algorithmic than the ordinary manual definition of the reference region, there are few operator-oriented uncertainties in AutoRef. We thus conclude that AutoRef can be applied as an automatic delineating algorithm for the reference region in amyloid imaging.

Four-dimensional fully convolutional residual network-based liver segmentation in Gd-EOB-DTPA-enhanced MRI.

PURPOSE: Gadolinium-ethoxybenzyl-diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced magnetic resonance imaging (MRI) tends to show higher diagnostic accuracy than other modalities. There is a demand for computer-assisted detection (CAD) software for Gd-EOB-DTPA-enhanced MRI. Segmentation with high accuracy is important for CAD software. We propose a liver segmentation method for Gd-EOB-DTPA-enhanced MRI that is based on a four-dimensional (4D) fully convolutional residual network (FC-ResNet). The aims of this study are to determine the best combination of an input image and output image in our proposed method and to compare our proposed method with the previous rule-based segmentation method. METHODS: We prepared a five-phase image set and a hepatobiliary phase image set as the input image sets to determine the best input image set. We also prepared a labeled liver image and labeled liver and labeled body trunk images as the output image sets to determine the best output image set. In addition, we optimized the hyperparameters of our proposed model. We used 30 cases to train our model, 10 cases to determine the hyperparameters of our model, and 20 cases to evaluate our model. RESULTS: Our network with the five-phase image set and the output image set of labeled liver and labeled body trunk images showed the highest accuracy. Our proposed method showed higher accuracy than the previous rule-based segmentation method. The Dice coefficient of the liver region was 0.944 ± 0.018. CONCLUSION: Our proposed 4D FC-ResNet showed satisfactory performance for liver segmentation as preprocessing in CAD software.

Can the spherical gold standards be used as an alternative to painted gold standards for the computerized detection of lesions using voxel-based classification?

PURPOSE: For the development of computer-assisted detection (CAD) software using voxel-based classification, gold standards defined by pixel-by-pixel painting, called painted gold standards, are desirable. However, for radiologists who define gold standards, a simplified method of definition is desirable. One of the simplest methods of defining gold standards is a spherical region, called a spherical gold standard. In this study, we investigated whether spherical gold standards can be used as an alternative to painted gold standards for computerized detection using voxel-based classification. MATERIALS AND METHODS: The spherical gold standards were determined by the center of gravity and the maximum diameter. We compared two types of gold standard, painted gold standards and spherical gold standards, by two types of CAD software using voxel-based classification. RESULTS: The time required to paint the area of one lesion was 4.7-6.5 times longer than the time required to define a spherical gold standard. For the same performance of the CAD software, the number of training cases required for the spherical gold standard was 1.6-7.6 times that for the painted gold standards. CONCLUSION: Spherical gold standards can be used as an alternative to painted gold standards for the computerized detection of lesions with simple shapes.

Deep neural network-based computer-assisted detection of cerebral aneurysms in MR angiography.

BACKGROUND: The usefulness of computer-assisted detection (CAD) for detecting cerebral aneurysms has been reported; therefore, the improved performance of CAD will help to detect cerebral aneurysms. PURPOSE: To develop a CAD system for intracranial aneurysms on unenhanced magnetic resonance angiography (MRA) images based on a deep convolutional neural network (CNN) and a maximum intensity projection (MIP) algorithm, and to demonstrate the usefulness of the system by training and evaluating it using a large dataset. STUDY TYPE: Retrospective study. SUBJECTS: There were 450 cases with intracranial aneurysms. The diagnoses of brain aneurysms were made on the basis of MRA, which was performed as part of a brain screening program. FIELD STRENGTH/SEQUENCE: Noncontrast-enhanced 3D time-of-flight (TOF) MRA on 3T MR scanners. ASSESSMENT: In our CAD, we used a CNN classifier that predicts whether each voxel is inside or outside aneurysms by inputting MIP images generated from a volume of interest (VOI) around the voxel. The CNN was trained in advance using manually inputted labels. We evaluated our method using 450 cases with intracranial aneurysms, 300 of which were used for training, 50 for parameter tuning, and 100 for the final evaluation. STATISTICAL TESTS: Free-response receiver operating characteristic (FROC) analysis. RESULTS: Our CAD system detected 94.2% (98/104) of aneurysms with 2.9 false positives per case (FPs/case). At a sensitivity of 70%, the number of FPs/case was 0.26. DATA CONCLUSION: We showed that the combination of a CNN and an MIP algorithm is useful for the detection of intracranial aneurysms. LEVEL OF EVIDENCE: 4 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2018;47:948-953.

Feasibility Study of a Generalized Framework for Developing Computer-Aided Detection Systems-a New Paradigm.

We propose a generalized framework for developing computer-aided detection (CADe) systems whose characteristics depend only on those of the training dataset. The purpose of this study is to show the feasibility of the framework. Two different CADe systems were experimentally developed by a prototype of the framework, but with different training datasets. The CADe systems include four components; preprocessing, candidate area extraction, candidate detection, and candidate classification. Four pretrained algorithms with dedicated optimization/setting methods corresponding to the respective components were prepared in advance. The pretrained algorithms were sequentially trained in the order of processing of the components. In this study, two different datasets, brain MRA with cerebral aneurysms and chest CT with lung nodules, were collected to develop two different types of CADe systems in the framework. The performances of the developed CADe systems were evaluated by threefold cross-validation. The CADe systems for detecting cerebral aneurysms in brain MRAs and for detecting lung nodules in chest CTs were successfully developed using the respective datasets. The framework was shown to be feasible by the successful development of the two different types of CADe systems. The feasibility of this framework shows promise for a new paradigm in the development of CADe systems: development of CADe systems without any lesion specific algorithm designing.

Automatic detection of vertebral number abnormalities in body CT images.

PURPOSE: The anatomical anomaly of the number of vertebral bones is one of the major anomalies in the human body, which can cause confusion of the spinal level in, for example, surgery. The aim of this study is to develop an automatic detection system for this type of anomaly. METHODS: We utilized our previously reported anatomical landmark detection system for this anomaly detection problem. This system uses a landmark point distribution model (L-PDM) to find multiple landmark positions. The L-PDM is a statistical probabilistic model of all landmark positions in the human body, including five landmarks for each vertebra. Given a new volume, the proposed algorithm applies five hypotheses (normal, 11 or 13 thoracic vertebrae, 4 or 6 lumbar vertebrae) to the given spine and attempts to detect all the landmarks. Then, the most plausible hypothesis with the largest posterior likelihood is selected as the anatomy detection result. RESULTS: The proposed method was evaluated using 300 neck-to-pelvis CT datasets. For normal subjects, the vertebrae of 211/217 (97.2%) of the subjects were successfully determined as normal. For subjects with 23 or 25 vertebrae without a transitional vertebra (TV), the vertebrae of 9/10 (90%) of the subjects were successfully determined. For subjects with TV, the vertebrae of 71/73 (97.3%) of subjects were judged as partially successfully determined. CONCLUSION: Our algorithm successfully determined the number of vertebrae, and the feasibility of our proposed system was validated.

Automatic detection of over 100 anatomical landmarks in medical CT images: A framework with independent detectors and combinatorial optimization.

An automatic detection method for 197 anatomically defined landmarks in computed tomography (CT) volumes is presented. The proposed method can handle missed landmarks caused by detection failure, a limited imaging range and other problems using a novel combinatorial optimization framework with a two-stage sampling algorithm. After a list of candidates is generated by each landmark detector, the best combination of candidates is searched for by a combinatorial optimization algorithm using a landmark point distribution model (L-PDM) to provide prior knowledge. Optimization is performed by simulated annealing and iterative Gibbs sampling. Prior to each cycle of Gibbs sampling, another sampling algorithm is processed to estimate the spatial distribution of each target landmark, so that landmark positions without any correct detector-derived candidates can be estimated. The proposed method was evaluated using 104 CT volumes with various imaging ranges. The overall average detection distance error was 6.6mm, and 83.8, 93.2 and 96.5% of landmarks were detected within 10, 15 and 20mm from the ground truth, respectively. The proposed method worked even when most of the landmarks were outside of the imaging range. The identification accuracy of the vertebral centroid was also evaluated using public datasets and the proposed method could identify 70% of vertebrae including severely diseased ones. From these results, the feasibility of our framework in detecting multiple landmarks in various CT datasets was validated.

Landmark-guided diffeomorphic demons algorithm and its application to automatic segmentation of the whole spine and pelvis in CT images.

PURPOSE: A fully automatic multiatlas-based method for segmentation of the spine and pelvis in a torso CT volume is proposed. A novel landmark-guided diffeomorphic demons algorithm is used to register a given CT image to multiple atlas volumes. This algorithm can utilize both grayscale image information and given landmark coordinate information optimally. METHODS: The segmentation has four steps. Firstly, 170 bony landmarks are detected in the given volume. Using these landmark positions, an atlas selection procedure is performed to reduce the computational cost of the following registration. Then the chosen atlas volumes are registered to the given CT image. Finally, voxelwise label voting is performed to determine the final segmentation result. RESULTS: The proposed method was evaluated using 50 torso CT datasets as well as the public SpineWeb dataset. As a result, a mean distance error of [Formula: see text] and a mean Dice coefficient of [Formula: see text] were achieved for the whole spine and the pelvic bones, which are competitive with other state-of-the-art methods. CONCLUSION: From the experimental results, the usefulness of the proposed segmentation method was validated.

Machine Learning for Computer-aided Diagnosis.

Machine learning algorithms are to analyze any dataset to extract data-driven model, prediction rule, or decision rule from the dataset. Various machine learning algorithms are now used to develop high-performance medical image processing systems such as computer-aided detection (CADe) system which detects clinically significant objects from medical images and computer-aided diagnosis (CADx) system which quantifies malignancy of manually or automatically detected clinical objects. In this paper, we introduce some applications of machine learning algorithms to the development of medical image processing system.

Development of automatic visceral fat volume calculation software for CT volume data.

OBJECTIVE: To develop automatic visceral fat volume calculation software for computed tomography (CT) volume data and to evaluate its feasibility. METHODS: A total of 24 sets of whole-body CT volume data and anthropometric measurements were obtained, with three sets for each of four BMI categories (under 20, 20 to 25, 25 to 30, and over 30) in both sexes. True visceral fat volumes were defined on the basis of manual segmentation of the whole-body CT volume data by an experienced radiologist. Software to automatically calculate visceral fat volumes was developed using a region segmentation technique based on morphological analysis with CT value threshold. Automatically calculated visceral fat volumes were evaluated in terms of the correlation coefficient with the true volumes and the error relative to the true volume. RESULTS: Automatic visceral fat volume calculation results of all 24 data sets were obtained successfully and the average calculation time was 252.7 seconds/case. The correlation coefficients between the true visceral fat volume and the automatically calculated visceral fat volume were over 0.999. CONCLUSIONS: The newly developed software is feasible for calculating visceral fat volumes in a reasonable time and was proved to have high accuracy.

Automatic categorization of anatomical landmark-local appearances based on diffeomorphic demons and spectral clustering for constructing detector ensembles.

A method for categorizing landmark-local appearances extracted from computed tomography (CT) datasets is presented. Anatomical landmarks in the human body inevitably have inter-individual variations that cause difficulty in automatic landmark detection processes. The goal of this study is to categorize subjects (i.e., training datasets) according to local shape variations of such a landmark so that each subgroup has less shape variation and thus the machine learning of each landmark detector is much easier. The similarity between each subject pair is measured based on the non-rigid registration result between them. These similarities are used by the spectral clustering process. After the clustering, all training datasets in each cluster, as well as synthesized intermediate images calculated from all subject-pairs in the cluster, are used to train the corresponding subgroup detector. All of these trained detectors compose a detector ensemble to detect the target landmark. Evaluation with clinical CT datasets showed great improvement in the detection performance.