Bone segmentation on whole-body CT using convolutional neural network with novel data augmentation techniques

Noguchi, Shunjiro; Nishio, Motohiro; Yakami, Masahiro; Nakagomi, Keita; Togashi, Kaori

doi:10.1016/j.compbiomed.2020.103767

Cited by 88 publications

(68 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…Table 4 shows that the combination of two types of data augmentation methods was more effective than single type or no data augmentation methods. Our results were consistent with the results of previous study done for bone segmentation with CNN 15 . Because the dataset of the current study was relatively small-sized (number of CXR images was 1248), it was necessary to improve the robustness of CNN models.…”

Section: Discussionsupporting

confidence: 93%

Automatic classification between COVID-19 pneumonia, non-COVID-19 pneumonia, and the healthy on chest X-ray image: combination of data augmentation methods

Nishio

Noguchi

Matsuo

et al. 2020

Sci Rep

Self Cite

119

View full text Add to dashboard Cite

This study aimed to develop and validate computer-aided diagnosis (CXDx) system for classification between COVID-19 pneumonia, non-COVID-19 pneumonia, and the healthy on chest X-ray (CXR) images. From two public datasets, 1248 CXR images were obtained, which included 215, 533, and 500 CXR images of COVID-19 pneumonia patients, non-COVID-19 pneumonia patients, and the healthy samples, respectively. The proposed CADx system utilized VGG16 as a pre-trained model and combination of conventional method and mixup as data augmentation methods. Other types of pre-trained models were compared with the VGG16-based model. Single type or no data augmentation methods were also evaluated. Splitting of training/validation/test sets was used when building and evaluating the CADx system. Three-category accuracy was evaluated for test set with 125 CXR images. The three-category accuracy of the CAD system was 83.6% between COVID-19 pneumonia, non-COVID-19 pneumonia, and the healthy. Sensitivity for COVID-19 pneumonia was more than 90%. The combination of conventional method and mixup was more useful than single type or no data augmentation method. In conclusion, this study was able to create an accurate CADx system for the 3-category classification. Source code of our CADx system is available as open source for COVID-19 research.

show abstract

Section: Discussionsupporting

confidence: 93%

Automatic classification between COVID-19 pneumonia, non-COVID-19 pneumonia, and the healthy on chest X-ray image: combination of data augmentation methods

Nishio

Noguchi

Matsuo

et al. 2020

Sci Rep

Self Cite

119

View full text Add to dashboard Cite

show abstract

“…This method could make up for the lack of data for the training, validation and testing of a CNN model, as this involves a set of altered images different from the original ones. In the case of data enhancement, this could solve the issue of insufficient data and improve the accuracy of convolutional neural network training [ 36 , 37 ].…”

Section: Materials and Methodologymentioning

confidence: 99%

Analyzing Malaria Disease Using Effective Deep Learning Approach

2020

View full text Add to dashboard Cite

Medical tools used to bolster decision-making by medical specialists who offer malaria treatment include image processing equipment and a computer-aided diagnostic system. Malaria images can be employed to identify and detect malaria using these methods, in order to monitor the symptoms of malaria patients, although there may be atypical cases that need more time for an assessment. This research used 7000 images of Xception, Inception-V3, ResNet-50, NasNetMobile, VGG-16 and AlexNet models for verification and analysis. These are prevalent models that classify the image precision and use a rotational method to improve the performance of validation and the training dataset with convolutional neural network models. Xception, using the state of the art activation function (Mish) and optimizer (Nadam), improved the effectiveness, as found by the outcomes of the convolutional neural model evaluation of these models for classifying the malaria disease from thin blood smear images. In terms of the performance, recall, accuracy, precision, and F1 measure, a combined score of 99.28% was achieved. Consequently, 10% of all non-dataset training and testing images were evaluated utilizing this pattern. Notable aspects for the improvement of a computer-aided diagnostic to produce an optimum malaria detection approach have been found, supported by a 98.86% accuracy level.

show abstract

“…It is worth mentioning that, although statistical shape modelling workflows present the advantage of reconstructing bone geometries from sparse segmentations or even skin landmarks, bone models from medical image segmentation still provide the most accurate estimations of joint parameters; for example, median root-mean-squared errors up to 11.09 mm and larger than 13.8 mm (Nolte et al, 2020) have been reported in the identification of the centre of the femoral head using statistical shape models. Considering that radiological scans are routinely collected to plan musculoskeletal surgical interventions and that the time required to segment bones (Noguchi et al, 2020) has decreased by orders of magnitudes thanks to recent deep learning techniques, the generation of personalised lower limb models in a number of clinical applications appears technically feasible.…”

Section: Discussionmentioning

confidence: 99%

Automatic Generation of Personalised Skeletal Models of the Lower Limb from Three-Dimensional Bone Geometries

Modenese

Renault

2020

Preprint

View full text Add to dashboard Cite

The generation of personalised and patient-specific musculoskeletal models is currently a cumbersome and time-consuming task that normally requires several processing hours and trained operators. We believe that this aspect discourages the use of computational models even when appropriate data are available and personalised biomechanical analysis would be beneficial. In this paper we present a computational tool that enables the fully automatic generation of skeletal models of the lower limb from three-dimensional bone geometries, normally obtained by segmentation of medical images. This tool was validated against four manually created lower limb models finding remarkable agreement in the computed joint parameters, well within human operator repeatability, with the only exception being the subtalar joint axis estimation, which was sensitive to the bone segmentation quality. To prove the robustness of the methodology, the models were built from datasets including both genders, anatomies ranging from juvenile to elderly and bone geometries reconstructed from high-quality computed tomography as well as lower-quality magnetic resonance imaging scans. The entire workflow, implemented in MATLAB scripting language, executed in seconds and required no operator intervention, creating lower extremity models ready to use for kinematic and kinetic analysis or as baselines for more advanced musculoskeletal modelling approaches, of which we provide some practical examples. We auspicate that this technical advancement will promote the use of personalised models in larger-scale studies than those hitherto undertaken.

show abstract

Bone segmentation on whole-body CT using convolutional neural network with novel data augmentation techniques

Cited by 88 publications

References 20 publications

Automatic classification between COVID-19 pneumonia, non-COVID-19 pneumonia, and the healthy on chest X-ray image: combination of data augmentation methods

Automatic classification between COVID-19 pneumonia, non-COVID-19 pneumonia, and the healthy on chest X-ray image: combination of data augmentation methods

Analyzing Malaria Disease Using Effective Deep Learning Approach

Automatic Generation of Personalised Skeletal Models of the Lower Limb from Three-Dimensional Bone Geometries

Contact Info

Product

Resources

About