Fig 1.
LSTM architecture gates: σ: Sigmoid activation function.
tanh: Hyperbolic tangent activation function, Operations: +: Addition, ×: Multiplication, States: X(t): input state, c(t): Current cell state, c(t-1): previous cel state, h(t): Current hidden state, h(t-1): previous hidden state.
Fig 2.
Bi-LSTM architecture.
Table 1.
Lung cancer detection based by utilizing optimized machine learning models and extracting hand-crafted GLCM features using 10-fold CV.
Fig 3.
K-fold cross validation method for training/ testing data formulation.
Fig 4.
AUC to distinguish SCLC from NSCLC utilizing different machine learning and deep learning algorithms a) SVM Linear, b) SVM Quadratic with 10 fold cross validation (CV).
Fig 5.
AUC to distinguish SCLC from NSCLC utilizing different machine learning and deep learning algorithms a) Decision Tree, b) Naïve Bayes, c) SVM Linear, d) SVM Quadratic, e) SVM Cubic, f) LSTM with 5 folds CV.
Table 2.
Lung cancer detection based by utilizing optimized machine learning models and extracting hand-crafted GLCM features using 5-fold CV.
Table 3.
Comparative analysis of various performance metrics for predicting lung cancer.
Fig 6.
Parallel Coordinate graph to distinguish NSCLC from SCLC by computing GLCM features from lung cancer using a) SVM Linear, b) SVM Cubic.
Table 4.
Lung cancer detection using optimized deep learning LSTM with 10-fold cross-validation.
Fig 7.
Accuracy-loss graph to detect Lung cancer using Bi-LSTM with different optimizers a) adam, rmsprop, c) sgdm.
Table 5.
Comparison of results with previous studies.