Computer-Vision-based-system-in-Matlab-for-Benign-and-Malignant-Classification-Skin-Lesion-Detection
Computer-Vision-based-system-in-Matlab-for-Benign-and-Malignant-Classification-Skin-Lesion-Detection
Review from Client ->
Required Method to be Implemented:
Phase 1: File Name: Skin_Lesion_Segmentation.m
Step 1: Set Segmentation Accuracy Limit (i.e 80% now).
Step 2: Run The code, it will automatically segment all the images from dataset path and will
save all images to either benign or malignant by using labels
provided in GroundTruth.txt file
Step 3: Command window will show the current progress of the segmentation algorithm.
Step 4: Figures and Subplot will show the visual performance of the segmentation algorithm.
Step 5: Finally, "000_Segmented.txt" file will be created having details about segmentation success cases.
Step 6: "000_FailCases.txt" file will be created having details for fail cases.
Phase 2: File Name: Skin_Lesion_Classification_Final.m
Step1: Set the Dataset portion Slpitting %
Step2: Set Feature Vector Reduction Ratio
Step3: Run the Code
Step4: After Fusion and Feature Vector Finalization, results can be evaluated using Classification Learner app from Computer Vision Tool Box.
Step 5: Save Classfier training File
Phase3:
FileName: Skin_Lesion_Classification_Final.m
Step 1 : Load the trained classifier from directory
Step 2 : Calculate the Sensitiy and Specificity using Confusion Matrix
Hog Visualization