Page 179 - Read Online
P. 179
Page 6 of 8 Arora et al. Neuroimmunol Neuroinflammation 2018;5:26 I http://dx.doi.org/10.20517/2347-8659.2018.11
Figure 3. Results depicting a benign tumor
morphology, segmentation and feature extraction using the sub-pixel edge detection algorithm and the
classification using the k-NN, a better observation of 71 accurate results in hundred iterations of testing was
given. This leads to an accuracy percentage of about 94.67% on the selected data set.
The above results are capable enough to ensure the operation of the system as a whole on any given dataset.
FUTURE ENHANCEMENTS
The speed of the system using DICOM images is still a challenge and we would want to focus on the DICOM
images in the further enhancements. The other motive will be to collect more images for both train and test
data sets and to have a better perspective on the testing of the system. The accuracy of the system can then
only be confirmed with the highest level of accuracy.