A CT-Based Radiomics Nomogram Model for Differentiating Primary Malignant Melanoma of the Esophagus from Esophageal Squamous Cell Carcinoma

Figure 4

The cases of ROI segmentation and CT-based radiomics models for diagnosing ESCC and PMME. (a–c) 60-year-old man with ESCC. (a) The ROI was drawn along the tumor boundary in plain CT; (b) plain CT and (c) enhanced CT. Radiomics signature with score of 0.22 (less than the cutoff value of 0.50) suggested the diagnosis of ESCC. The radiomics nomogram model showed the radiomics score of 0.22 corresponded to 28 points, and the probability of diagnosing PMME was less than 0.01 indicating the diagnosis of ESCC. (d–f) 46-year-old women with PMME. (d) The ROI was drawn along the tumor boundary in plain CT; (e) plain CT and (f) enhanced CT. Radiomics signature with score of 0.98 (larger than the cutoff value of 0.50) suggested the diagnosis of PMME. The radiomics nomogram model showed the radiomics score of 0.98 corresponded to 98 points, and the probability of diagnosing PMME was larger than 0.99 stating the diagnosis of PMME.

(a)

(b)

(c)

(d)

(e)

(f)