ECIDS-Enhanced Cancer Image Diagnosis and Segmentation Using Artificial Neural Networks and Active Contour Modelling

Abstract

In the present decade, image processing techniques are extensively utilized in various medical image diagnoses, specifically in dealing with cancer images for detection and treatment in advance. The quality of the image and the accuracy are the significant factors to be considered while analyzing the images for cancer diagnosis. With that note, in this paper, an Enhanced Cancer Image Diagnosis and Segmentation (ECIDS) framework has been developed for effective detection and segmentation of lung cancer cells. Initially, the Computed Tomography lung image (CT image) has been processed for denoising by employing kernel based global denoising function. Following that, the noise free lung images are given for feature extraction. The images are further classified into normal and abnormal classes using Feed Forward Artificial Neural Network Classification. With that, the classified lung cancer images are given for segmentation and the process of segmentation has been done here with the Active Contour Modelling with reduced gradient. The segmented cancer images are further given for medical processing. Moreover, the framework is experimented with MATLAB tool using the clinical dataset called LIDC-IDRI lung CT dataset. The results are analyzed and discussed based on some performance evaluation metrics such as energy, Entropy, Correlation and Homogeneity are involved in effective classification.

Lung Tumor Staging and Classification with Machine Learning and Deep Learning Models

“…The paper [40] utilized wavelet transform based feature extraction to extract the texture features of lung images. In a different manner, Genetic Algorithm (GA) has been enforced in [19], [41] for cancer diagnosis. However, the feature detection process is complicated with the number of CT images creates time complexity.…”