CNN-based deep learning method for predicting the disease response of NAC (Neoadjuvant Chemotherapy) treatment method in Breast Cancer

Abstract

Abstract Healthcare institutions produce big heterogeneous data in different structures and sources daily. Depending on this situation, the prediction of being able to make sense of and manage the data in this structure with traditional methods could decrease. Depending on this situation, the performance of interpreting and managing data in this structure decreases with traditional methods. It is a powerful tool for managing, interpreting, and analyzing such data with machine learning and deep learning methods. The correct diagnosis of the disease and the correct analysis of pathological data depend on obtaining and interpreting the appropriate data for prediction. NAC (Neoadjuvant Chemotherapy), a treatment method used in breast cancer cases discussed in the study, aimed to predict patients' response to treatment and the disease development process in the pathological area. Classification performances of CNN-based proposed models for tumor status after NAC treatment have been evaluated in detail through pathological data frequently used in the healthcare industry. The number of convolutional layers, data set quality, and main criteria that may affect the model's success during training have been evaluated. Since it can offer strong feature representation, results have been obtained with scenarios based on CNN models from deep neural networks. Interpreting the pathological test results with deep learning methods in determining the correct diagnosis and treatment method with the prognosis follow-up of the patient provides clinicians with a solution to a large extent.