当前位置: X-MOL 学术Curr. Comput.-Aided Drug Des. › 论文详情
Our official English website, www.x-mol.net, welcomes your feedback! (Note: you will need to create a separate account there.)
Computer-Aided Detection System for the Classification of Non-Small Cell Lung Lesions using SVM
Current Computer-Aided Drug Design ( IF 1.7 ) Pub Date : 2020-11-30 , DOI: 10.2174/1573409916666200102122021
Shruti Jain 1
Affiliation  

Introduction: Lung carcinoma is the most commonly cancer causing deaths throughout the world that mainly occurs due to smoking. Small cell lung cancer and Non-small cell lung cancer (NSCLC) are the two different types of Lung cancer. For the detection and classification of lung cancer, there are different techniques in the literature.

Methods: This paper emphasis on the three class classification of the Adenocarcinomas, Squamous cell carcinomas, and large cell carcinomas of NSCLC. For precise and superior results, Computer Aided Detection (CADe) system has been designed so that the radiologist can diagnose carcinoma in the ultrasonic images conveniently. CADe analyses the quality of the images, selects the region of interest, preprocesses the data, extracts the features and classifies the cancer.

Results: After exhaustive literature survey, Laws’ mask features and SVM classifier with Gaussian RBF kernels have been used in this paper. The experimentation was performed on 92 images using 50% - 50% training and testing criteria.

Conclusion: Comparative study reveals that our system for separating three class lung cancer provides 95.65% average accuracy for Laws' mask 3 dimensions using the SVM classifier that is maximum among the existing methods reported in the literature using the same dataset.



中文翻译:

使用 SVM 对非小细胞肺病变进行分类的计算机辅助检测系统

简介:肺癌是全世界最常见的导致死亡的癌症,主要由吸烟引起。小细胞肺癌和非小细胞肺癌 (NSCLC) 是两种不同类型的肺癌。对于肺癌的检测和分类,文献中有不同的技术。

方法:本文重点介绍非小细胞肺癌的腺癌、鳞状细胞癌和大细胞癌的三类分类。为了获得精确和卓越的结果,设计了计算机辅助检测 (CADe) 系统,以便放射科医生可以方便地诊断超声图像中的癌症。CADe 分析图像质量、选择感兴趣区域、预处理数据、提取特征并对癌症进行分类。

结果:经过详尽的文献调查,本文使用了 Laws 的掩码特征和具有高斯 RBF 核的 SVM 分类器。使用 50% - 50% 的训练和测试标准对 92 张图像进行了实验。

结论:比较研究表明,我们用于分离三类肺癌的系统使用 SVM 分类器为 Laws 掩码 3 维提供了 95.65% 的平均准确率,这是使用相同数据集的文献中报道的现有方法中最高的。

更新日期:2021-01-19
down
wechat
bug