Enhanced pre-processing technique for histopathological image stain normalization and cancer detection

Cancer is a significant global public health issue, with lung, colon, and breast cancers being among the most common and deadliest, together accounting for over 25% of all cases. Early detection is crucial for improving survival rates. This paper presents a classification framework for differentiating types of lung and colon tissues and breast cancer by analyzing histopathological images using modern Artificial Intelligence (AI), Machine Learning (ML), and Digital Image Processing techniques. The study utilizes the BreakHis and LC25000 datasets for breast, colon, and lung cancer. The methodology includes comprehensive image preprocessing, feature extraction, and classification. Initial noise reduction is achieved using the Quadruple Clipped Adaptive Histogram Equalization technique, and the Stain Color Adaptive Normalization (SCAN) algorithm is applied to preserve local characteristics and optimize contrast. An Arithmetic Optimization Algorithm is developed to enhance SCAN parameters and improve the machine learning model's performance. The VGG16 model is employed for feature extraction, followed by an Ensemble max-voting classifier for classification. The proposed method significantly advances cancer detection in medical image analysis, resulting in quicker and more precise diagnoses. The experimental results demonstrate a classification accuracy of 89.03%, precision of 88.56%, sensitivity of 88.09%, F1-score of 88.7%, specificity of 89.08%, PCC of 0.894, SSIM of 0.944, NMI of 0.898, and rSM of 0.956.