Deep Learning Techniques for Pancreatic Cancer Analysis: A Systematic Review and Implantation Prerequisites

Abstract

Listen

Translate article

Of all the cases of pancreatic cancer reported in 2022, 52% of patients belonged to stage IV with a 2.9% 5-year survival rate. Lower survival rates are reported because the advanced-stage incidences leave only <20% of patients eligible for in-patient tumor resect-ability. Due to the fatal consequences, researchers have been rigorously implementing artificial intelligence techniques for early detection and analysis of cancer. This systematic review scrutinizes published research from 2020 to 2023 to map key concepts based on objectives, techniques, imaging modalities, datasets, and current limitations in the emerging field of deep learning strategies applied to image analysis of the ailment. To achieve this aim, a research methodology was developed to investigate articles pertinent to deep learning deployment, radiological imaging modalities, and publicly available imaging datasets for pancreatic cancer classification. Based on the findings, the authors discussed major applications, the maximum applied computed tomography imaging modality, and 3 datasets. The authors emphasized future enhancements including multimodal detection, dataset generation, and ways to improve clinical applications. The summarized results also enlist optimum implementation prerequisites for the disease’s detection which are substantial to aid researchers, scholars, and academicians in gaining deep insights into related strategies for future experiments.