| Sumario: | A deep learning based robust classification is presented using Deep SPAE (Stacked Patch Auto-Encoder) optimized using Black-Widow Optimization (BWO) algorithm in anatomical structures which is commonly called as Deep SPAE_BWO. Initially, pre-processing is done using GIF (Guided Image Filter) eliminates the noise and preserves the sharp edges. Next normalization using Min_Max method makes changes in the range of pixel values. Then, the image structures which possess similar characteristics are grouped with Modified C- Mean (MC-M) clustering. Feature Extraction is done using Entropy based Local Binary Pattern (ELBP), Grey Level Co-occurrence Matrix (GLCM) and Grey Level Run Length Matrix (GLRLM) to extract the optimal features. The tissues may be incompletely visible and may look different across images. So, Deep SPAE_BWObased classification is used to classify the anatomical structures followed by automatic confidence estimation named Gini-co-efficient (GC). Finally, an automatic image tagging (IT) approach is presented to label the classified images as Liver, Gallbladder and Fat Tissues. This work is implemented in PYTHON tool. The classification performance is evaluated in terms of precision, specificity, accuracy, sensitivity, NPV (Negative Predictive Value) and F1-score.
|
|---|