Thyroid Cancer Gene Detection Algorithm Based on Feature Selection ELM
Wang Jining, HaoYang Mi, Yubo Wu, Xingtong Li, Chaohui Lin · 2016
At present, the detection and diagnosis of thyroid cancer has always been depend on the puncture cytology of thyroid gland.However, this method (the puncture cytology of thyroid gland) demands high accuracy of the instrument and costs may stay at a relative high level.Under this circumstance, aiming at solving this problem, our team comes up with a method, which is the thyroid cancer gene detection algorithm based on feature, to assist medical detection.The experiment of 100 genetic samples from DDBJ human gene bank shows that, the thyroid cancer gene detection algorithm based on feature selection ELM method can effectively improve the result of the thyroid cancer detection.Thyroid cancer is one of the most common types of cancer, if the patients are not cured at an early-stage, once developed to end-stage, the cancer will lead to lifelong medication or life-threatening.In addition, some of the complications of thyroid cancer such as the damage to liver and kidney functions, which are seriously affected the daily life of patients.But now medical diagnostic tool for thyroid cancer diagnosis can not be 100%, so the supplementary medical means to help diagnose thyroid cancer are needed.The current means of computer-aided diagnosis of thyroid cancer associated with gene expression data is the use of the R language for DNA microarray data to identify and to give the results of computer-aided diagnosis.The literature describes the feasibility of using DNA microarray data for gene selection, classification and machine learning.Finally achieve the feasibility of auxiliary diagnosis.But the disadvantage of this approach is that the success rate is not high, after the completion of the diagnosis, a lot of conventional physical and chemical examination are still required to diagnose patients whether suffering from thyroid cancer or not.To solve this problem, this article proposes some methods to increase the success rate: A. Using the correlation analysis to filter the7 data present in the DNA microarray with genes involved in thyroid cancer screening; B. Proposed thyroid cancer diagnosis method based on gene expression data of ELM; C. The use of the data downloaded from the NCBI demonstrates the feasibility of the method described above and obtained a better diagnosis. Research BackgroundComputer Aided Diagnosis is a medical approach through the using the computer imaging, image processing technology and gene expression data to improve the diagnosis rate.Classification of genetic samples plays an extremely important part in computer-aided diagnosis of gene expression data detection field, effective classification can simplify the genetic data operation, and post-classification using machine learning methods to speed up the diagnostic process, shortening