Validation of a Multitask Deep Learning System at Spine Metastasis CT

Multicenter Validation of a Multitask Deep Learning System for Spine Metastases Detection Andfeatures Evaluations at Spine Metastasis CT

15 resident oncologists were conducted to evaluate the clinical efficacy of DLS in multicenter. They were 1:1 randomly asked to independently read the test images without the assistance of DLS software or with the assistance. Area under the receiver operating characteristic curve (AUC), sensitivity, and specificity of the DLS were calculated with professional graders as the reference standard.

Inclusion Criteria:

1. pathology-proven diagnosis of solid tumor;
2. spinal CT scan indicating spinal metastasis with at least one lesion;
3. no previous surgery for spinal metastasis

Exclusion Criteria:

1. spinal CT scans with no sagittal reconstruction;
2. the radiologist considered that the quality of CT image was unqualified.