Deep neural network with generative adversarial networks pre-training for brain tumor classification based on MR images

Publication date: March 2020Source: Biomedical Signal Processing and Control, Volume 57Author(s): Navid Ghassemi, Afshin Shoeibi, Modjtaba RouhaniAbstractIn this paper, a new deep learning method for tumor classification in MR images is presented. A deep neural network is first pre-trained as a discriminator in a generative adversarial network (GAN) on different datasets of MR images to extract robust features and to learn the structure of MR images in its convolutional layers. Then the fully connected layers are replaced and the whole deep network is trained as a classifier to distinguish three tumor classes. The deep neural network classifier has six layers and about 1.7 million weight parameters. Pre-training as a discriminator of a GAN together with other techniques such as data augmentations (image rotation and mirroring) and dropout prevent the network from overtraining on a relatively small dataset. This method is applied to an MRI data set consists of 3064 T1-CE MR images from 233 patients, 13 images from each patient on average, with three different brain tumor types: meningioma (708 images), glioma (1426 images), and pituitary tumor (930 images). 5-Fold cross-validation is used to evaluate the performance of overall design, achieving the highest accuracy as compared to state-of-art methods.
Source: Biomedical Signal Processing and Control - Category: Biomedical Science Source Type: research

Related Links:

Publication date: Available online 10 December 2019Source: Artificial Intelligence in MedicineAuthor(s): Sarah Ali Abdelaziz Ismael, Ammar Mohammed, Hesham HefnyAbstractCancer is the second leading cause of death after cardiovascular diseases. Out of all types of cancer, brain cancer has the lowest survival rate. Brain tumors can have different types depending on their shape, texture, and location. Proper diagnosis of the tumor type enables the doctor to make the correct treatment choice and help save the patient's life. There is a high need in the Artificial Intelligence field for a Computer Assisted Diagnosis (CAD) syste...
Source: Artificial Intelligence in Medicine - Category: Bioinformatics Source Type: research
Conclusions This review describes how leukocyte-heparanase can be a double-edged sword in tumor progression; it can enhance tumor immune surveillance and tumor cell clearance, but also promote tumor survival and growth. We also discuss the potential of using heparanase in leukocyte therapies against tumors, and the effects of heparanase inhibitors on tumor progression and immunity. We are just beginning to understand the influence of heparanase on a pro/anti-tumor immune response, and there are still many questions to answer. How do the pro/anti-tumorigenic effects of heparanase differ across different cancer types? Does...
Source: Frontiers in Oncology - Category: Cancer & Oncology Source Type: research
Ramya Raman, Anantanarayanan RamanIndian Journal of Cancer 2018 55(4):424-427 A three-page paper entitled “A case of glioma (embryonal neurocytoma) of the brain simulating pituitary tumour,” written by K. G. Pandalai, Surgeon at the Madras General Hospital, and T. B. Menon, Pathologist at the Madras Medical College, was published in the Indian Medical Gazette. This paper refers to an occasional finding of a brain tumor (a glioma), presurgically diagnosed as a pituitary tumor, in 1928. This was close to the period when surgeons across the developed world believed in that cancers were by no means common among th...
Source: Indian Journal of Cancer - Category: Cancer & Oncology Authors: Source Type: research
Publication date: Available online 18 October 2018Source: Biocybernetics and Biomedical EngineeringAuthor(s): Amin Kabir Anaraki, Moosa Ayati, Foad KazemiAbstractGliomas are the most common type of primary brain tumors in adults and their early detection is of great importance. In this paper, a method based on convolutional neural networks (CNNs) and genetic algorithm (GA) is proposed in order to noninvasively classify different grades of Glioma using magnetic resonance imaging (MRI). In the proposed method, the architecture (structure) of the CNN is evolved using GA, unlike existing methods of selecting a deep neural netw...
Source: Biocybernetics and Biomedical Engineering - Category: Biomedical Engineering Source Type: research
AbstractBrain tumors are associated with increased mortality and morbidity and are the most common cancer type in children and young adults. The present review focuses on the interplay between leptin, the most extensively studied adipokine, and the onset, development, and treatment of primary brain and intracranial tumors. The two main mechanisms for increased leptin levels in intracranial tumor survivors, leptin resistance caused by hypothalamic damage, or secondary to obesity, are discussed. The contradicting mechanistic observations on leptin being able to both promote tumorinogenesis (e.g., in gliomas) as well as inhib...
Source: Hormones and Cancer - Category: Cancer & Oncology Source Type: research
In conclusion, these results provide compelling evidence that miR-520d-5p functions as an anti-onco-miRNA, which is important in inhibiting cell proliferation in GBM, and its anti-oncogenic effects are mediated chiefly through direct suppression of PTTG1 expression. Therefore, we suggest that miR-520d-5p is a potential candidate for the prevention of glioblastoma. PMID: 29218086 [PubMed]
Source: American Journal of Translational Research - Category: Research Tags: Am J Transl Res Source Type: research
This study aimed to characterize a subset of patients at highest risk for new-onset perioperative seizures (i.e., intraoperative and postoperative seizures occurring within 30 days of surgery) who may benefit from prophylactic AEDs. METHODS The authors conducted a retrospective case-control study of all adults who had undergone tumor resection or biopsy at the authors' institution between January 1, 2004, and June 31, 2015. All patients with a history of preoperative seizures, posterior fossa tumors, pituitary tumors, and parasellar tumors were excluded. A control group was matched to the seizure patients according to age ...
Source: Journal of Neurosurgery - Category: Neurosurgery Authors: Tags: J Neurosurg Source Type: research
by Jun Cheng, Wei Yang, Meiyan Huang, Wei Huang, Jun Jiang, Yujia Zhou, Ru Yang, Jie Zhao, Yanqiu Feng, Qianjin Feng, Wufan Chen Content-based image retrieval (CBIR) techniques have currently gained increasing popularity in the medical field because they can use numerous and valuable archived images to support clinical decisions. In this paper, we concentrate on developing a CBIR system for retrieving brain tumors in T1-weighted contrast-enhanced MRI images. Specifically, when the user roughly outlines the tumor region of a query image, brain tumor images in the database of the same pathological type are expected to be re...
Source: PLoS One - Category: Biomedical Science Authors: Source Type: research
Examination of tissue sections using desorption electrospray ionization (DESI)-MS revealed phospholipid-derived signals that differ between gray matter, white matter, gliomas, meningiomas, and pituitary tumors, allowing their ready discrimination by multivariate statistics. A set of lower mass signals, some corresponding to oncometabolites, including 2-hydroxyglutaric acid and N-acetyl-aspartic acid, was also observed...
Source: Proceedings of the National Academy of Sciences - Category: Science Authors: Tags: Inaugural Articles Physical Sciences Source Type: research
by Jun Cheng, Wei Huang, Shuangliang Cao, Ru Yang, Wei Yang, Zhaoqiang Yun, Zhijian Wang, Qianjin Feng Automatic classification of tissue types of region of interest (ROI) plays an important role in computer-aided diagnosis. In the current study, we focus on the classification of three types of brain tumors (i.e., meningioma, glioma, and pituitary tumor) in T1-weighted contrast-enhanced MRI (CE-MRI) images. Spatial pyramid matching (SPM), which splits the image into increasingly fine rectangular subregions and computes histograms of local features from each subregion, exhibits excellent results for natural scene classific...
Source: PLoS One - Category: Biomedical Science Authors: Source Type: research
More News: Biomedical Science | Brain | Brain Cancers | Brain Tumor | Glioma | Learning | Men | Neurology | Pituitary Tumor | Training | Universities & Medical Training