AI-Assisted Digital Pathology for Computational distinction of adenocarcinoma vs squamous cell carcinoma

Lung diseases owing to their high morbidity and mortality pose a significant burden to the healthcare systems. The evolution in integrated circuit technology and the consequent rise in computational power has paved the way for the realization of digital pathology (DP) tools. These DP tools equipped with recent developments in artificial intelligence (AI) can assist pathologists in improving clinical evaluations and thereby prescribing betterinformed treatments. Previously researchers have proposed such DP tools for the treatments of breast, prostate, and head and neck cancers, however, developing such tools and applying them for the treatment of lung diseases remains a challenge. The applications of AI-based DP tools on radiology images for improved characterization of lung nodules, lung cancer risk stratification, and fibrotic lung disease are well established. However, the subtypes of non-small cell lung cancer (NSCLC), namely, adenocarcinomas (AC) and squamous cell carcinoma (SCC) are still difficult to distinguish, especially in small biopsies. In this article, we discuss potential architectures of convolutional neural networks (CNN) for the construction of an appropriate AI-based DP classification tool in establishing a crisp distinction between adenocarcinoma and squamous cell carcinoma. It is further emphasized that the construction of such a DP tool will help in mitigating the current issues that the pathologists are facing while working with the conventional molecular biomarkers profiling, a process which is both, expensive and requires high technical and interpretative skills.