A Computed Tomography-based Radiomics Model for Patient Prognostic Stratification in Non-Small Cell Lung Cancer
Mitchell Chen, Yidong Han, Cillian McNamara, Marize Bakhet, Christos Argyros, William Crum, Eric O. Aboagye, Susan J Copley
Purpose/Objectives
To develop and validate a prognostic predictive vector using a comprehensive radiomic feature analysis of CTs for patients with non-small-cell lung cancer (NSCLC).
Methods and Materials
In this retrospective, ethically approved study, 947 patients with non-small cell lung cancer diagnosed at our multicenter institution between 2009 and 2018 were included. Patient overall survival at 3 years was used as the primary study endpoint. The three centres (A, B, C) at our institution used varying scanners and scanning protocols and thus formed separate internal cohorts, with the largest (A) used as our discovery cohort. Patients who had received surgery with a curative intent to their primary cancer formed a subgroup, due to more favourable expected prognosis. The discovery cohort was split 2:1 into training and internal validation sets, balanced for patient's age, sex, and tumour histology. Our external testing set consisted of publicly available data from the Cancer Imaging Archive.
Two trained radiologists, blinded to clinical and histological data, with 13 and 6 years of chest imaging experience respectively, double reviewed all scans and performed semi-automated segmentation of the tumour, peritumoral penumbra, and a spherical parenchymal patch in the surrounding pulmonary lobe. Key radiological findings such as pleural attachment, effusion, and tumour margin spiculation are documented.
The imaging data were comprehensively analysed for a total of 1998 radiomic using in-house software. Unsupervised hierarchical clustering based on Euclidean distance was performed for each histological subtype separately on the normalised radiomic profile. Supervised learning was achieved by least absolute shrinkage and selection operation regularisation following univariate Cox regression on each radiomic feature, to derive a composite radiomic prognostic vector (RPV) model.
Results
Unsupervised hierarchical clustering (shown in Figure 1) achieved statistically significant difference in survival between the groups for both histology subtypes (adenocarcinoma: p < 0.0001, squamous cell: p = 0.043), and enrichment for various tumour stage and radiological finding metrics, as shown in Figure 2.
The model was successfully validated on the internal validation, internal and external testing cohorts (p-value 0.00473, Site B: < 0.0001, Site C: 0.00013, Surgical: 0.0489, TCIA: 0.025, respectively), as shown in Figure 3. Superior performance of RPV was demonstrated over standalone volumetry in all four testing cohorts (volumetry-only predictive model p-value: Site B: 0.0002, Site C: 0.0306, Surgical: 0.235, and TCIA: 0.149).
Conclusion
Our work shows the clear benefit of incorporating advanced radiomic features into a clinical predictive model for NSCLC patients, for improving personalised cancer care.

**A Computed Tomography-based Radiomics Model for Patient Prognostic Stratification in Non-Small Cell Lung Cancer**
Mitchell Chen, Yidong Han, Cillian McNamara, Marize Bakhet, Christos Argyros, William Crum, Eric O. Aboagye, Susan J Copley
Purpose/Objectives
To develop and validate a prognostic predictive vector using a comprehensive radiomic feature analysis of CTs for patients with non-small-cell lung cancer (NSCLC).
Methods and Materials
In this retrospective, ethically approved study, 947 patients with non-small cell lung cancer diagnosed at our multicenter institution between 2009 and 2018 were included. Patient overall survival at 3 years was used as the primary study endpoint. The three centres (A, B, C) at our institution used varying scanners and scanning protocols and thus formed separate internal cohorts, with the largest (A) used as our discovery cohort. Patients who had received surgery with a curative intent to their primary cancer formed a subgroup, due to more favourable expected prognosis. The discovery cohort was split 2:1 into training and internal validation sets, balanced for patient's age, sex, and tumour histology. Our external testing set consisted of publicly available data from the Cancer Imaging Archive.
Two trained radiologists, blinded to clinical and histological data, with 13 and 6 years of chest imaging experience respectively, double reviewed all scans and performed semi-automated segmentation of the tumour, peritumoral penumbra, and a spherical parenchymal patch in the surrounding pulmonary lobe. Key radiological findings such as pleural attachment, effusion, and tumour margin spiculation are documented.
The imaging data were comprehensively analysed for a total of 1998 radiomic using in-house software. Unsupervised hierarchical clustering based on Euclidean distance was performed for each histological subtype separately on the normalised radiomic profile. Supervised learning was achieved by least absolute shrinkage and selection operation regularisation following univariate Cox regression on each radiomic feature, to derive a composite radiomic prognostic vector (RPV) model.
Results
Unsupervised hierarchical clustering (shown in Figure 1) achieved statistically significant difference in survival between the groups for both histology subtypes (adenocarcinoma: p < 0.0001, squamous cell: p = 0.043), and enrichment for various tumour stage and radiological finding metrics, as shown in Figure 2.
The model was successfully validated on the internal validation, internal and external testing cohorts (p-value 0.00473, Site B: < 0.0001, Site C: 0.00013, Surgical: 0.0489, TCIA: 0.025, respectively), as shown in Figure 3. Superior performance of RPV was demonstrated over standalone volumetry in all four testing cohorts (volumetry-only predictive model p-value: Site B: 0.0002, Site C: 0.0306, Surgical: 0.235, and TCIA: 0.149).
Conclusion
Our work shows the clear benefit of incorporating advanced radiomic features into a clinical predictive model for NSCLC patients, for improving personalised cancer care.

A Computed Tomography-based Radiomics Model for Patient Prognostic Stratification in Non-Small Cell Lung Cancer

The Winter Course 2021 covers a variety of topics from basics in interstitial lung disease, infections including Covid-19, malignancies, lung cancer screening, and artificial intelligence. The combination of topics is aimed to serve as a good introduction to radiologists embarking on a career in thoracic imaging as well as a solid refresher for experienced thoracic radiologists.

[ESTI Winter Course 2021 Evaluation Form](https://www.surveymonkey.com/r/PP9WFTJ?sid=%5Bsid_value%5D&uid=%5Buid_value%5D&tkn=%5Btkn_value%5D&cgrid=%5Bcgrid_value)

ESTI Winter Course 2021

Break Video

CT-Guided Microcoil placement for VATS Resection of Pulmonary Nodules

Development of a risk prediction model for pneumothorax in percutaneous CT guided lung biopsy

Evaluation of an AI program in thoracic imaging: Performance in detection of thoracic aorta ectasia

A Single Centre Institutional Study on 2295 patients for COVID-19 Detection from Radiographs Using a Deep Learning Model - Our Experience and Insights

Evaluation of AI applications in thoracic imaging: Performance in the detection of vertebral body alterations

Can inclusion of the combination of chemical shift ratio and signal intensity increase the number of patient’s classified as definitively benign on thymic MRI beyond use of chemical shift ratio alone? Experience from a UK thoracic tertiary referral centre

Relative contribution of emphysema and airway disease to disease severity in COPD patients GOLD stages I to IV using quantitative computed tomography

Lung Thermal Ablation: Comparison between an Augmented Reality Computed Tomography (CT) 3D Navigation System (SIRIO) and Standard CT-Guided Technique

Differentiating COVID-19 pneumonia from atypical bacterial, fungal and non-COVID-19 viral pneumonia on Chest-CT: Is it possible?

Reliability of Ultra-Short Echo Time MRI sequence in the assessment of Interstitial Lung Disease related to Systemic Sclerosis

Assessing longitudinal changes in lung gas transfer in dyspnoeic post-COVID-19 patients using hyperpolarized Xenon-129 Magnetic Resonance Imaging

Session Introduction