SEGMENTATION OF LUNG CANCER IMAGES WITH THRESHOLD TECHNIQUE COMPARED WITH K-MEANS CLUSTERING

Aim: The main aim of the study is lung tumor segmentation, which is used to detect a tumor in lung CT images with the proper segmentation method. In this research work, Threshold segmentation is compared with the k-mean clustering method. The lung CT image is enhanced with a Gabor filter and is processed for segmentation. Materials and methods: A total of 178 different lung CT image samples is analysed from the dataset available in Github. These samples are tested based on the tumor area. Threshold segmentation and k-mean cluster segmentation are used in the segmentation process of lung cancer tumors. The sample size was calculated by using Clincalc.com. Results: From the experimental and statistical analysis it is observed that threshold segmentation achieved high accuracy of 94.56%, while k-mean cluster segmentation gives an accuracy of 87.62%. with the significant value of p=0.21, hence the two groups are not significantly different. Conclusion: In this study,threshold segmentation appears to give better results when compared to the k-mean clustering.