• Journal of Biomedical Engineering Research
• /
• v.41 no.2
• /
• pp.101-106
• /
• 2020

Although benign gastric ulcers do not develop into gastric cancer, they are similar to early gastric cancer and difficult to distinguish. This may lead to misconsider early gastric cancer as gastric ulcer while diagnosing. Since gastric cancer does not have any special symptoms until discovered, it is important to detect gastric ulcers by early gastroscopy to prevent the gastric cancer. Therefore, we developed a Convolution Neural Network (CNN) model that can be helpful for endoscopy. 3,015 images of gastroscopy of patients undergoing endoscopy at Gachon University Gil Hospital were used in this study. Using ResNet-50, three models were developed to classify normal and gastric ulcers, normal and gastric cancer, and gastric ulcer and gastric cancer. We applied the data augmentation technique to increase the number of training data and examined the effect on accuracy by varying the multiples. The accuracy of each model with the highest performance are as follows. The accuracy of normal and gastric ulcer classification model was 95.11% when the data were increased 15 times, the accuracy of normal and gastric cancer classification model was 98.28% when 15 times increased likewise, and 5 times increased data in gastric ulcer and gastric cancer classification model yielded 87.89%. We will collect additional specific shape of gastric ulcer and cancer data and will apply various image processing techniques for visual enhancement. Models that classify normal and lesion, which showed relatively high accuracy, will be re-learned through optimal parameter search.

• Asian Pacific Journal of Cancer Prevention
• /
• v.17 no.2
• /
• pp.835-838
• /
• 2016

Background: Lung cancer remains one of the most common cancers in the world, both in terms of new cases (about 13% of total per year) and deaths (nearly one cancer death in five), because of the high case fatality. Errors in lung cancer type or malignant growth determination lead to degraded treatment efficacy, because anticancer strategy depends on tumor morphology. Materials and Methods: We have made an attempt to evaluate effectiveness of machine learning algorithms in the task of lung cancer classification based on gene expression levels. We processed four publicly available data sets. The Dana-Farber Cancer Institute data set contains 203 samples and the task was to classify four cancer types and sound tissue samples. With the University of Michigan data set of 96 samples, the task was to execute a binary classification of adenocarcinoma and non-neoplastic tissues. The University of Toronto data set contains 39 samples and the task was to detect recurrence, while with the Brigham and Women's Hospital data set of 181 samples it was to make a binary classification of malignant pleural mesothelioma and adenocarcinoma. We used the k-nearest neighbor algorithm (k=1, k=5, k=10), naive Bayes classifier with assumption of both a normal distribution of attributes and a distribution through histograms, support vector machine and C4.5 decision tree. Effectiveness of machine learning algorithms was evaluated with the Matthews correlation coefficient. Results: The support vector machine method showed best results among data sets from the Dana-Farber Cancer Institute and Brigham and Women's Hospital. All algorithms with the exception of the C4.5 decision tree showed maximum potential effectiveness in the University of Michigan data set. However, the C4.5 decision tree showed best results for the University of Toronto data set. Conclusions: Machine learning algorithms can be used for lung cancer morphology classification and similar tasks based on gene expression level evaluation.

• Asian Pacific Journal of Cancer Prevention
• /
• v.17 no.5
• /
• pp.2375-2382
• /
• 2016

Cancer is a major health problem in Oman. It is reported that cancer incidence in Oman is the second highest after Saudi Arabia among Gulf Cooperation Council countries. Based on GLOBOCAN estimates, Oman is predicted to face an almost two-fold increase in cancer incidence in the period 2008-2020. However, cancer research in Oman is still in its infancy. This is due to the fact that medical institutions and infrastructure that play central roles in data collection and analysis are relatively new developments in Oman. We believe the country requires an organized plan and efforts to promote local cancer research. In this paper, we discuss current research progress in cancer diagnosis using machine learning techniques to optimize computer aided cancer detection and classification (CAD). We specifically discuss CAD using two major medical data, i.e., medical imaging and microarray gene expression profiling, because medical imaging like mammography, MRI, and PET have been widely used in Oman for assisting radiologists in early cancer diagnosis and microarray data have been proven to be a reliable source for differential diagnosis. We also discuss future cancer research directions and benefits to Oman economy for entering the cancer research and treatment business as it is a multi-billion dollar industry worldwide.

• Journal of Gastric Cancer
• /
• v.15 no.3
• /
• pp.183-190
• /
• 2015

Purpose: The Lauren classification system is a very commonly used pathological classification system of gastric adenocarcinoma. A recent study proposed that the Lauren classification should be modified to include the anatomical location of the tumor. The resulting three types were found to differ significantly in terms of genomic expression profiles. This retrospective cohort study aimed to evaluate the clinical significance of the modified Lauren classification (MLC). Materials and Methods: A total of 677 consecutive patients who underwent curative gastrectomy from January 2005 to December 2007 for histologically confirmed gastric cancer were included. The patients were divided according to the MLC into proximal non-diffuse (PND), diffuse (D), and distal non-diffuse (DND) type. The groups were compared in terms of clinical features and overall survival. Multivariate analysis served to assess the association between MLC and prognosis. Results: Of the 677 patients, 48, 358, and 271 had PND, D, and DND, respectively. Their 5-year overall survival rates were 77.1%, 77.7%, and 90.4%. Compared to D and PND, DND was associated with significantly better overall survival (both P<0.01). Multivariate analysis showed that age, differentiation, lympho-vascular invasion, T and N stage, but not MLC, were independent prognostic factors for overall survival. Multivariate analysis of early gastric cancer patients showed that MLC was an independent prognostic factor for overall survival (odds ratio, 5.946; 95% confidence intervals, 1.524~23.197; P=0.010). Conclusions: MLC is prognostic for survival in patients with gastric adenocarcinoma, in early gastric cancer. DND was associated with an improved prognosis compared to PND or D.

• Genomics & Informatics
• /
• v.18 no.4
• /
• pp.35.1-35.7
• /
• 2020

Identifying the patterns of gene expression in breast cancers is essential to understanding their pathophysiology and developing anticancer drugs. Breast cancer is a heterogeneous disease with different subtypes determined by distinct biological features. Luminal breast cancer is characterized by a relatively high expression of estrogen receptor (ER) and progesterone receptor (PR) genes, which are expressed in breast luminal cells. In ~25% of invasive breast cancers, human epidermal growth factor receptor 2 (HER2) is overexpressed; these cancers are categorized as the HER2 type. Triple-negative breast cancer (TNBC), in which the cancer cells do not express ER/PR or HER2, shows highly aggressive clinical outcomes. TNBC can be further classified into specific subtypes according to genomic mutations and cancer immunogenicity. Herein, we discuss the brief history of TNBC classification and its implications for promising treatments.

• Safety and Health at Work
• /
• v.14 no.2
• /
• pp.141-152
• /
• 2023

Objective: We performed a meta-analysis of epidemiological results for the association between occupational exposure as a firefighter and cancer as part of the broader evidence synthesis work of the IARC Monographs program. Methods: A systematic literature search was conducted to identify cohort studies of firefighters followed for cancer incidence and mortality. Studies were evaluated for the influence of key biases on results. Random-effects meta-analysis models were used to estimate the association between ever-employment and duration of employment as a firefighter and risk of 12 selected cancers. The impact of bias was explored in sensitivity analyses. Results: Among the 16 included cancer incidence studies, the estimated meta-rate ratio, 95% confidence interval (CI), and heterogeneity statistic (I²) for ever-employment as a career firefighter compared mostly to general populations were 1.58 (1.14-2.20, 8%) for mesothelioma, 1.16 (1.08-1.26, 0%) for bladder cancer, 1.21 (1.12-1.32, 81%) for prostate cancer, 1.37 (1.03-1.82, 56%) for testicular cancer, 1.19 (1.07-1.32, 37%) for colon cancer, 1.36 (1.15-1.62, 83%) for melanoma, 1.12 (1.01-1.25, 0%) for non-Hodgkin lymphoma, 1.28 (1.02-1.61, 40%) for thyroid cancer, and 1.09 (0.92-1.29, 55%) for kidney cancer. Ever-employment as a firefighter was not positively associated with lung, nervous system, or stomach cancer. Results for mesothelioma and bladder cancer exhibited low heterogeneity and were largely robust across sensitivity analyses. Conclusions: There is epidemiological evidence to support a causal relationship between occupational exposure as a firefighter and certain cancers. Challenges persist in the body of evidence related to the quality of exposure assessment, confounding, and medical surveillance bias.

• Journal of KIISE:Computer Systems and Theory
• /
• v.35 no.8
• /
• pp.372-377
• /
• 2008

Multiclass cancer classification has been actively investigated based on gene expression profiles, where it determines the type of cancer by analyzing the large amount of gene expression data collected by the DNA microarray technology. Since gene expression data include many genes not related to a target cancer, it is required to select informative genes in order to obtain highly accurate classification. Conventional rank-based gene selection methods often use ideal marker genes basically devised for binary classification, so it is difficult to directly apply them to multiclass classification. In this paper, we propose a novel method for multiclass gene selection, which does not use ideal marker genes but directly analyzes the distribution of gene expression. It measures the class-discriminability by discretizing gene expression levels into several regions and analyzing the frequency of training samples for each region, and then classifies samples by using the naive Bayes classifier. We have demonstrated the usefulness of the proposed method for various representative benchmark datasets of multiclass cancer classification.

• Asian Pacific Journal of Cancer Prevention
• /
• v.16 no.9
• /
• pp.4109-4114
• /
• 2015

Background: Full-field digital mammography (FFDM) with dense breasts has a high rate of missed diagnosis, and digital breast tomosynthesis (DBT) could reduce organization overlapping and provide more reliable images for BI-RADS classification. This study aims to explore application of COMBO (FFDM+DBT) for effect and significance of BI-RADS classification of breast cancer. Materials and Methods: In this study, we selected 832 patients who had been treated from May 2013 to November 2013. Classify FFDM and COMBO examination according to BI-RADS separately and compare the differences for glands in the image of the same patient in judgment, mass characteristics display and indirect signs. Employ Paired Wilcoxon rank sum test was used in 79 breast cancer patients to find differences between two examine methods. Results: The results indicated that COMBO pattern is able to observe more details in distribution of glands when estimating content. Paired Wilcoxon rank sum test showed that overall classification level of COMBO is higher significantly compared to FFDM to BI-RADS diagnosis and classification of breast (P<0.05). The area under FFDM ROC curve is 0.805, while that is 0.941 in COMBO pattern. COMBO shows relation of mass with the surrounding tissues, the calcification in the mass, and multiple foci clearly in breast cancer tissues. The optimal sensitivity of cut-off value in COMBO pattern is 82.9%, which is higher than that in FFDM (60%). They share the same specificity which is both 93.2%. Conclusions: Digital Breast Tomosynthesis (DBT) could be used for the BI-RADS classification in breast cancer in clinical.

• Proceedings of the Korean Institute of Information and Commucation Sciences Conference
• /
• 2021.10a
• /
• pp.655-657
• /
• 2021

It is challenging to find breast ultrasound image training dataset to develop an accurate machine learning model due to various regulations, personal information issues, and expensiveness of acquiring the images. However, studies targeting transfer learning for ultrasound breast cancer images classification have not been able to achieve high performance compared to radiologists. Here, we propose an improved transfer learning model for ultrasound breast cancer classification using publicly available dataset. We argue that with a proper combination of ImageNet pre-trained model and optimizer, a better performing model for ultrasound breast cancer image classification can be achieved. The proposed model provided a preliminary test accuracy of 99.5%. With more experiments involving various hyperparameters, the model is expected to achieve higher performance when subjected to new instances.

• International Journal of Computer Science & Network Security
• /
• v.21 no.6
• /
• pp.101-106
• /
• 2021

Breast cancer is the number one cause of deaths from cancer in women, knowing the type of breast cancer in the early stages can help us to prevent the dangers of the next stage. The performance of the deep learning depends on large number of labeled data, this paper presented convolutional neural network for classification breast cancer from images to benign or malignant. our network contains 11 layers and ends with softmax for the output, the experiments result using public BreakHis dataset, and the proposed methods outperformed the state-of-the-art methods.