A Review on Pathology Report based Cancer Diagnosing System using Intelligent Techniques

Abstract

OBJECTIVE: The aim of this paper is to review the significant advances made in the context of extracting information from clinical pathology reports, with an emphasis on reports pertaining to cancer.

METHOD: The development of a fast growing body of research focused on applying machine learning and, more recently, deep learning to cancer diagnosis and to the specific domain of extraction of cancer-related information from pathology reports is the motivation for this survey. We have reviewed a large number of relevant research papers in this field and in related domains, with a focus on most recent research, and evaluated them from the viewpoint of progress towards automated cancer surveillance.

RESULTS: It is observed that deep learning techniques have significantly outperformed traditional machine learning approaches, as evident from latest research in this field. However, machine learning and NLP techniques are equally important, for high accuracy results as well as to understand the evolution of research trends in the field. The potential of deep learning applications remains to be fully utilized in this domain to achieve enhanced cancer surveillance