Segmentation of Non-small Cell Lung Cancer (NSCLC) Segmentation on CT Images
| Main Authors: | Mohamed Afiq Akmal Mohamed Akil, Siti Salasiah Mokri, Ashrani Aizuddin Abd Rahni, Asraf Mohamed Moubark, Mohd Hairi Mohd Zaman |
|---|---|
| Format: | Proceeding eJournal |
| Bahasa: | eng |
| Terbitan: |
, 2019
|
| Subjects: | |
| Online Access: |
https://zenodo.org/record/3356398 |
Daftar Isi:
- Segmentation is an important step in medical diagnosis applications which affects the accuracy of the overall system. The goal of this project is to determine the most appropriate Non Small Cell Lung Cancer (NSCLC) segmentation method on CT images in which 5 patient datasets were used archived from http://michallenges.org/moistRun. Comparison between 3 segmentation methods was done, namely multilevel thresholding, region growing, and level set. The region of interest (ROI) was manually defined using either free hand or cubic mask. To evaluate the segmentation accuracy, the segmented structures were examined with respect to the ground truth segmentation results according to Jaccard coefficient, Dice coefficient, False Positive Ratio (RFP) and False Negative Ratio (RFN). The results showed that level set segmentation method that used freehand ROI definition achieved the best segmentation performance as compared the other two methods.