Table 1.
Clinical characteristics of study patients.
Fig 1.
CT image of lymph node metastasis from small cell lung cancer.
A large lymph node, which was proven by a transbronchial needle aspirate to be a lymph node metastasis of small cell lung cancer, is depicted on an enhanced CT image (A). Density and contrast enhancement of the node appears slightly heterogeneous; however, massive necrosis is not observed (B).
Fig 2.
CT image of enlarged lymph nodes due to sarcoidosis.
An enlarged mediastinal lymph node, as well as slightly enlarged hilar nodes, are shown on an enhanced CT image (A). The density of the mediastinal lymph node seems basically homogeneous (B); however, distinguishing enlargement due to sarcoidosis from that due to lymph node metastasis from small cell lung cancer on CT images only is difficult.
Fig 3.
An example of texture analysis performed by the software.
The open-source software (LIFEx) performed texture analysis of a 10-mm-diamter circular region of interest (colored pink), which was placed in the center of the targeted lymph node.
Fig 4.
Receiver Operating Characteristic (ROC) curves of the values for CT-based texture features and US elastographic FLRs.
ROC curves of the grey-level run length matrix (GLRLM)-high gray-level emphasis (HGRE) feature and the fat-to-lesion strain ratio (FLR), and the ROC of the combination of GLRLM-HGRE and FLR, which shows almost perfect diagnostic accuracy for distinguishing between lymphadenopathy due to sarcoidosis and that due to metastasis from small cell lung cancer. The area-under-the-curve values were as follows: GLRLM-HGRE alone, 0.781; FLR alone, 0.857; combination of GLRLM-HGRE plus FLR, 0.988.
Table 2.
Radiomics features of lymph nodes affected by sarcoidosis versus the features of lymph nodes affected by small cell carcinoma.