Design and Development of A Diagnostic System for Early Prediction of P53 Mutation Causing Cancer from Microscopic Biopsy Images

Abstract

The major complication associated with cancer care is delayed cancer detection, which would also reduce the likelihood of survival. This situation could be resolved to some extend with an early diagnostic system. In the current study, designing an early detection system for TP53 mutation, which is a common primary mutation for most of the types of cancer, has been carried out using the ‘Pharmacogenomics’, ‘Gene expression profiling’ and ‘Deep imaging processing technique’. The input for the analysis is microscopic biopsy images collected from the ‘Expression atlas database’. The high level of expression of TP53 gene mutation has been observed in Breast and Ovarian cancers samples. The involvement of associated genes like BARD1, CHEK2, ATM, BRCA2, BRCA1, and RAD51 has also been analyzed. A deep neural network with a ‘Siamese Neural Network (SNN)’, architecture has been implemented using one-short learning process to comprehend the data and make valid predictions on TP53 mutation. This ‘algorithm and learning platform’ helps in making dependable predictions even from a low input data and the machine's measured predictive performance is 89%.