Please enable JavaScript.

Coggle requires JavaScript to display documents.

Lung Abnormality, Bhandary, A., Prabhu, A., Rajinikanth, V., Krishnan, P.,…

- - - - Clinical-grade CT scans
      - Used 777 slices
  - - - Separates lung sections from rest of image
  - - - Extraction of essential features (18)
      - Nodule related features (9)
- - - - CPU constraint
  - - - Precision = 96.58%
      - Sensitivity = 96.97%
      - Accuracy = 96.80%
      - Specificity = 96.63%
      - F1 Score = 96.78%
- - - - CT Scan
      - X-Ray
      - MRI
    - - Time-consuming
      - Laborious
  - - - CT Scan most common
    - - Feature extraction using
        
        separation of high and low contrast regions
        
        Use of morphological filters
        
        RGB to Grayscale
        
        Tumor segmentation
        
        Extraction of the Harlick and Hu features
- - - - Addition of Convolution, ReLU, normalization, and max-pooling
      - Improves feature extraction
      - Softmax is replaced by SVM with linear kernel
  - - - Concatenation of deep features
      - Hand-crafted are also concatenated
      - Use of MAN architecture
      - Principal Component Analysis