Comprehensive Perspective for Lung Cancer Characterisation Based on AI Solutions Using CT Images

Abstract

Lung cancer is still the leading cause of cancer death in the world. For this reason, novel approaches for early and more accurate diagnosis are needed. Computer-aided decision (CAD) can be an interesting option for a noninvasive tumour characterisation based on thoracic computed tomography (CT) image analysis. Until now, radiomics have been focused on tumour features analysis, and have not considered the information on other lung structures that can have relevant features for tumour genotype classification, especially for epidermal growth factor receptor (EGFR), which is the mutation with the most successful targeted therapies. With this perspective paper, we aim to explore a comprehensive analysis of the need to combine the information from tumours with other lung structures for the next generation of CADs, which could create a high impact on targeted therapies and personalised medicine. The forthcoming artificial intelligence (AI)-based approaches for lung cancer assessment should be able to make a holistic analysis, capturing information from pathological processes involved in cancer development. The powerful and interpretable AI models allow us to identify novel biomarkers of cancer development, contributing to new insights about the pathological processes, and making a more accurate diagnosis to help in the treatment plan selection.

Machine Learning and Feature Selection Methods for EGFR Mutation Status Prediction in Lung Cancer

“…Nevertheless, the results of the current study indicate that the CT features with the highest correlation with EGFR mutation are from the lung that has the nodule and these are therefore the main contributors to the model decision. It is crucial to highlight these results and further investigate the importance of holistic lung cancer characterization studies, as there are many complex combinations of morphological, molecular, and genetic alterations that occur during lung cancer development that, when taken into account, would allow the development of more accurate classifiers for EGFR mutation status prediction [21].…”

“…Generally, the nodule is the main focus for lung cancer malignancy assessment and follow-up based on the well-established Fleischner Society and Lung-RADs Guidelines [17,18], and previous studies have analyzed only this important cluster of tumor cells for EGFR mutation status prediction [10,14,16]. Exploratory studies have recently shown that there is a correlation of EGFR mutation status with other lung diseases, such as emphysema and fibrosis, which seems to indicate that cancer development is related to multiple physiological changes not restricted to the nodule region [19,20], and that the inclusion of extratumor features allows a significant increase in EGFR mutation predictive performance [4,5,21].…”

Towards Machine Learning-Aided Lung Cancer Clinical Routines: Approaches and Open Challenges

“…This structure must be detected, segmented, and assessed to make the initial diagnosis of malignancy. For the malignant cases, CADs could help in cancer characterization, based on more information from the lung structures surrounding the nodules since other lung pathologies are correlated with cancer development [ 69 ].…”

“…Exploratory studies that have taken into account features from multiple lung structures, and did not just focus on the nodule, showed the importance of including extra-tumor features to obtain a successful genomic prediction (see Figure 3 , where it is illustrated that radiogenomics approaches use information from more than just the nodule region) [ 50 , 69 , 70 , 71 , 72 , 73 ]. This seems to indicate that cancer development is related to multiple physiological changes not restricted to the nodule region and that the next generation of CADs should consider large lung regions to allow for a more complete lung cancer characterization [ 16 , 74 ].…”

Multi-Input Dual-Stream Capsule Network for Improved Lung and Colon Cancer Classification

“…Deep learning algorithms for classifying and diagnosing lung and colon cancer using histopathology images have become a popular research topic in recent years [39], however, due to a paucity of data, no substantial progress has been achieved so far [40]. Despite the lack of data, a few authors have contributed significantly [41].…”