Breast Cancer Detection using ML.
Machine learning (ML) is widely recognized as the methodology of choice in cancer pattern classification because of its unique advantages in critical feature detection from complex data sets. Cancer is one of the biggest issues. According to statistics, every year there are 18 million new cases and 9.5 million cancer-related deaths worldwide. Breast cancer contributes to around 27% of all cancers. According to the World Health Organization, breast cancer was responsible for 502,000 deaths in 2005 alone and 1,301,867 new cases of breast cancer resisted. The discussion will be based on the image-based dataset Breast Cancer Detection using ML. Convolutional Neural Network is a type of neural network technique that has proven to be specifically efficient with image recognition and classification.
Block Diagram
There can be different types of input datasets for training the model including text, image data. Steps included in design of the model.
1. Data exploration
The absence of exact data regarding the classification of the tumor into benign and malignant in case of scan images of patients. Generally, issues regarding glitches in the training process leading to the improper organization of data and generation of huge redundant records. The research about the available dataset that can be used for the project is an important decision. As a consequence, it can be concluded that evaluation based totally on detection rate and accuracy levels is not appropriate.
2. Selection of new dataset if required
Hence it is appropriate for the training and testing part of the project which classifies the image into benign and malignant.
3. Pre-Processing of data
A data mining technique that is used to filter data in a usable format is data preprocessing. Since the real-world dataset is almost available in different formats. It's not available as per our requirement so it must fit the dataset in an understandable format. A proven method to resolve such issues can be attained using data preprocessing. Conversion of the dataset into a usable format for pre-processing is performed by data pre-processing. Standardization and Normalization methods are used to preprocess the dataset.
4. Training and Testing
The input is a tensor with shape (number of images) x (image width) x (image height) x (image depth) in the programming CNN. Then the image becomes abstracted to a feature map, with shape (number of images) x (feature map width) x (feature map height) x (feature map channels) after passing through a convolutional layer. Concatenating the datasets for processing performance is required since the training and testing datasets are separate files in the source. First, the unique values in each nonnumerical column are displayed to understand how encoding should be done. For encoding, the categorical data Label Encoder can be used. The following attributes are expected in a convolutional layer within a neural network.
Width and height (hyper-parameters) defined by convolutional kernels.
The number of input channels and output channels also known as hyper-parameters.
The number channels (depth) of the input feature map must be equal to the depth of the Convolution filter (the input channels).
ReLU is the abbreviation of the rectified linear unit, which applies the non-saturating activation function. It successfully removes negative values from an activation map by setting them to zero. The nonlinear properties of the overall network and the decision function without affecting the receptive fields of the convolution layer are increased by it.
Neural Network Functionality
5. Accuracy
Learning the pattern of the data and generalizing well for the unseen data in order to predict the result is the main objective of the ML model. The accuracy of the model is a very important aspect of the model since medical analysis is based on these outcomes. It should be as high as possible for reliable conclusions.
Conclusion
There are many other possibilities and methods to approach cancer detection using machine learning. Using images as training data is one of them. Many researchers are into the biomedical field and are leveraging diverse technologies such as Machine learning, Cloud computing, Big data, and many more. Hope you enjoyed learning!
Comentários