Khan Ayesha Adil, Narejo Ghous Bakhsh
Department of Electronics Engineering, NED University of Engineering & Technology, Karachi, Pakistan.
Curr Med Imaging Rev. 2019;15(10):972-982. doi: 10.2174/1573405615666190716122040.
The application of image processing algorithms for medical image analysis has been found effectual in the past years. Imaging techniques provide assistance to the radiologists and physicians for the diagnosis of abnormalities in different organs.
The proposed algorithm is designed for automatic computer-aided diagnosis of liver cancer from low contrast CT images. The idea expressed in this article is to classify the malignancy of the liver tumor ahead of liver segmentation and to locate HCC burden on the liver.
A novel Fuzzy Linguistic Constant (FLC) is designed for image enhancement. To classify the enhanced liver image as cancerous or non-cancerous, fuzzy membership function is applied. The extracted features are assessed for malignancy and benignancy using the structural similarity index. The malignant CT image is further processed for automatic tumor segmentation and grading by applying morphological image processing techniques.
The validity of the concept is verified on a dataset of 179 clinical cases which consist of 98 benign and 81 malignant liver tumors. Classification accuracy of 98.3% is achieved by Support Vector Machine (SVM). The proposed method has the ability to automatically segment the tumor with an improved detection rate of 78% and a precision value of 0.6.
The algorithm design offers an efficient tool to the radiologist in classifying the malignant cases from benign cases. The CAD system allows automatic segmentation of tumor and locates tumor burden on the liver. The methodology adopted can aid medical practitioners in tumor diagnosis and surgery planning.
在过去几年中，已发现将图像处理算法应用于医学图像分析是有效的。成像技术为放射科医生和内科医生诊断不同器官的异常情况提供了帮助。
所提出的算法旨在用于从低对比度CT图像中自动进行肝癌的计算机辅助诊断。本文所表达的想法是在肝脏分割之前对肝肿瘤的恶性程度进行分类，并确定肝脏上肝癌的负荷。
设计了一种新颖的模糊语言常数（FLC）用于图像增强。为了将增强后的肝脏图像分类为癌性或非癌性，应用了模糊隶属函数。使用结构相似性指数评估提取的特征的恶性和良性程度。通过应用形态学图像处理技术，对恶性CT图像进一步进行自动肿瘤分割和分级。
在由98个良性和81个恶性肝肿瘤组成的179个临床病例的数据集上验证了该概念的有效性。支持向量机（SVM）实现了98.3%的分类准确率。所提出的方法能够自动分割肿瘤，检测率提高到78%，精确值为0.6。
该算法设计为放射科医生提供了一种有效的工具，用于将恶性病例与良性病例区分开来。计算机辅助诊断（CAD）系统允许自动分割肿瘤并确定肝脏上的肿瘤负荷。所采用的方法可以帮助医生进行肿瘤诊断和手术规划。
