br Fig shows the comparison
Fig. 22 shows the comparison of three samples of CT images which undergo morphological operation, ABM conventional, ABM with corner feature and ABM with local minima-maxima features. Red region, green line and white area represent the location of nod-ule, the border of the Angiotensin II obtained by using lung extraction step and the covering area of each method, respectively.
The results show inability of ABM min-max in detecting small juxta pleural nodules. All methods obtain US values of more than 10% because there are many juxta-pleural or juxta-vascular nod-ules that cannot be covered by the methods. An example of this case is illustrated in Fig. 23.
Moreover, the values of OS in all methods are more than 100%. The large values of OS occur because the vascular areas have some basins which have similar characteristics with nodule basin. This condition is illustrated in Fig. 20(a). All aforementioned methods
Fig. 23. One example of the hard nodule is covered by aforementioned methods; (a) original image, (b) morphological, (c) ABM-conventional, (d) ABM-corner, and (e) ABM-minmax.
Fig. 24. The segmentation results of lung (a) without boundary correction, (b) with boundary correction.
are able to cover the juxta-pleural and juxta-vascular nodules in order to improve the result of Otsu thresholding method. The 3D illustration result of the covered nodule basin obtained by the pro-posed method is shown in Fig. 24.
The objectives of ABM-modified methods are to reduce compu-tational time and to reduce FP. By using the features represented nodule basin, the ABM-modified methods successfully reduce computational time of more 60 times than that of ABM conven-tional. The proposed ABM-corner method in this study achieves the fastest computational time of 0.32 s/slice. According to the results, the proposed method obtains the best performance of under segmentation with 14.6% in comparison to ABM min-max and morphological based methods.
In summary, ABM-corner has better performance to maintain the small nodule basin than that of ABM min–max. It is because ABM-corner is more able to detect the curvature in all direction than the ABM with local minima and maxima. However, the effect of sensitivity in detecting small nodule basin is large number of FP. Finding the optimum rules to detect small nodule basin with low FP and FP reduction are still necessary for further study.
Conflict of interest
The authors declared that there is no ‘Conflict of interest’.
The authors would like to acknowledge the Department of Elec-trical and Information Engineering, Universitas Gadjah Mada and Directorate General of Higher Education, Ministry of Research, Technology and Higher Education, Republic of Indonesia for fund-ing Back mutation research work through the ‘‘Penelitian Tim Pasca Sarjana” Research Grant. The authors would also like to thank colleagues of Intelligent System research group in our Department for inspiring discussion and anonymous reviewers for encouraging reviews and recommendations.
American Cancer Society, Cancer Facts & Figures 2017. Atlanta: American Cancer Society, 2017.
Dhara, A.K., Mukhopadhyay, S., Khandelwal, N., 2012. Computer-aided detection and analysis of pulmonary nodule from CT Images: a survey. IETE Tech. Rev. 29
Gupta, A., Martens, O., Le Moullec, Y., 2015. Methods for increased sensitivity and scope in automatic segmentation and detection of lung nodules in CT images. In: 2015 IEEE International Symposium on Signal Processing and Information Technology (ISSPIT), vol. 2, pp. 375–380.
Please cite this article as: R. Nurfauzi, H. A. Nugroho, I. Ardiyanto et al., Autocorrection of lung boundary on 3D CT lung cancer imagesq, Journal of King Saud University – Computer and Information Sciences, https://doi.org/10.1016/j.jksuci.2019.02.009
10 R. Nurfauzi et al. / Journal of King Saud University – Computer and Information Sciences xxx (xxxx) xxx
‘‘LIDC-IDRI - The Cancer Imaging Archive (TCIA) Public Access - Cancer Imaging Archive Wiki.” [Online]. Available: https://wiki.cancerimagingarchive.net/ display/Public/LIDC-IDRI (accessed: 21-Feb-2017).
M.R.S., 2016. Automatic 3D pulmonary nodule detection in CT Images: a survey.
Zhou, S., Cheng, Y., Tamura, S., 2014. Biomedical signal processing and control automated lung segmentation and smoothing techniques for inclusion of juxtapleural nodules and pulmonary vessels on chest CT images. Biomed. Signal Process. Control 13, 62–70.
Contents lists available at ScienceDirect
Expert Systems With Applications