Geometric Artifact Evaluation of X-ray Computed Tomography Images Based on Convolutional Neural Network

Author(s):  
Mingwan Zhu ◽  
Linlin Zhu ◽  
Yu Han ◽  
Xiaoqi Xi ◽  
Lei Li ◽  
...  
2021 ◽  
Vol 9 (B) ◽  
pp. 1283-1289
Author(s):  
Jane Aurelia ◽  
Zuherman Rustam

BACKGROUND: Cancer is a major health problem not only in Indonesia but also throughout the world. Cancer is the growth and spread of abnormal cells that have distinctive characteristics, that if can no longer be controlled will usually cause death. The number of deaths due to cancer is generally caused by late diagnosis and inappropriate treatment. To reduce mortality from cancer, it is necessary to strive for early detection and monitoring of cancer in patients undergoing therapy. Convolutional neural networks (CNNs) as one of machine learning methods are designed to produce or process data from two dimensions that have a network tier and many applications carried out in the image. Moreover, support vector machines (SVMs) as a hypothetical space in the form of linear functions feature have high dimensions and trained algorithm based on optimization theory. AIM: In connection with the above, this paper discusses the role of the machine learning technique named a hybrid CNN-SVM. METHODS: The proposed method is used in the detection and monitoring of cancers by determining the classification of cancers in X-ray computed tomography (CT) patients’ images. Several types of cancer that used for determination in detection and monitoring of cancers diagnosis are also discussed in this paper, such as lung, liver, and breast cancer. RESULTS: From the discussion, the results show that the combining model of hybrid CNN-SVM has the best performance with 99.17% accuracy value. CONCLUSION: Therefore, it can be concluded that machine learning plays a very important role in the detection and management of cancer treatment through the determination of classification of cancers in X-ray CT patients’ images. As the proposed method can detect cancer cells with an effective mechanism of action so can has the potential to inhibit in the future studies with more extensive data materials and various diseases.


2021 ◽  
Vol 20 ◽  
pp. 153303382110101
Author(s):  
Thet-Thet Lwin ◽  
Akio Yoneyama ◽  
Hiroko Maruyama ◽  
Tohoru Takeda

Phase-contrast synchrotron-based X-ray imaging using an X-ray interferometer provides high sensitivity and high spatial resolution, and it has the ability to depict the fine morphological structures of biological soft tissues, including tumors. In this study, we quantitatively compared phase-contrast synchrotron-based X-ray computed tomography images and images of histopathological hematoxylin-eosin-stained sections of spontaneously occurring rat testicular tumors that contained different types of cells. The absolute densities measured on the phase-contrast synchrotron-based X-ray computed tomography images correlated well with the densities of the nuclear chromatin in the histological images, thereby demonstrating the ability of phase-contrast synchrotron-based X-ray imaging using an X-ray interferometer to reliably identify the characteristics of cancer cells within solid soft tissue tumors. In addition, 3-dimensional synchrotron-based phase-contrast X-ray computed tomography enables screening for different structures within tumors, such as solid, cystic, and fibrous tissues, and blood clots, from any direction and with a spatial resolution down to 26 μm. Thus, phase-contrast synchrotron-based X-ray imaging using an X-ray interferometer shows potential for being useful in preclinical cancer research by providing the ability to depict the characteristics of tumor cells and by offering 3-dimensional information capabilities.


Author(s):  
Ilya Straumit ◽  
Christoph Hahn ◽  
Elisabeth Winterstein ◽  
Bernhard Plank ◽  
Stepan V. Lomov ◽  
...  

2019 ◽  
Vol 69 (3) ◽  
pp. 185-187
Author(s):  
Magnus Fredriksson ◽  
Julie Cool ◽  
Stavros Avramidis

Abstract X-ray computed tomography (CT) scanning of sawmill logs is associated with costly and complex machines. An alternative scanning solution was developed, but its data have not been evaluated regarding detection of internal features. In this exploratory study, a knot detection algorithm was applied to images of four logs to evaluate its performance in terms of knot position and size. The results were a detection rate of 67 percent, accurate position, and inaccurate size. Although the sample size was small, it was concluded that automatic knot detection in coarse resolution CT images of softwoods is feasible, albeit for knots of sufficient size.


2015 ◽  
Vol 2 (2) ◽  
pp. 104-117 ◽  
Author(s):  
Masaji Kato ◽  
Manabu Takahashi ◽  
Satoru Kawasaki ◽  
Katsuhiko Kaneko

Sign in / Sign up

Export Citation Format

Share Document