Detection of Pleuro Pulmonary Blastoma using Machine Learning Models

Abstract

Pleura Pulmonary Blastoma (PPB) is a type of lung cancer seen in children. PPB needs to be detected earlier when treating children. The mortality rate of PPB is higher if left untreated. It can be detected from CT images through various machine learning and classification algorithms. The earlier detection of PPB can save children's lives, for which several research works have proposed several machine learning models. Several researchers adopt traditional classification algorithms like random forest and decision tree algorithms for detecting PPB. However, these techniques provided lesser accuracy and were difficult for earlier detections. This paper considered several machine learning algorithms like SVM, LR and MP and experimented with CT images and DICER-1 data to understand their betterness and overcome such issues. The architecture of the following algorithms is discussed in detail, and the results are compared. Through this, the ideal machine learning algorithm for detecting PPB is found. All the algorithms are implemented with the Python software, and the performance metrics of the respective algorithms are recorded. The results show that the SVM algorithm provides better accuracy (96%) for the DICER-1 dataset, which is higher than CT images (95.60%).