Imaging Phenotyping Using Radiomics to Predict Micropapillary Pattern within Lung Adenocarcinoma

Area of Science:

• Oncology
• Radiology
• Medical Imaging

Background:

• Lung adenocarcinoma (ADC) with a micropapillary pattern is associated with poor prognosis.
• Objective imaging identification of this pattern is limited, relying on subjective qualitative CT variables.

Purpose of the Study:

• To explore radiomics for noninvasive prediction of the micropapillary pattern in lung ADC.
• To assess the value of imaging phenotyping in identifying aggressive tumor components.

Main Methods:

• Retrospective analysis of 339 lung ADC patients.
• Histological classification and quantification of micropapillary component.
• Assessment of clinical features and conventional imaging variables.
• Quantitative CT analysis using radiomics features (histogram, texture, shape).

Main Results:

• Higher tumor stage and intermediate grade predicted a micropapillary component.
• Lower minimum pixel value and lower variance of positive pixel value were predictive.
• Conventional imaging metrics (max SUV, TDR) did not significantly differentiate groups.

Conclusions:

• Radiomics offers a noninvasive method to interrogate tumors for micropapillary patterns.
• Combining radiomics with clinical features enhances diagnostic value for treatment planning.