Use of systems with deep learning and machine learning for the detection and classification of malignant vs. benign spinal fractures with MRI: can deep/machine learning help us further for detection and characterization?

Excerpt

In the first study [1], an artificial intelligence (AI)–based system with deep learning (DL) was developed to automatically differentiate malignant/benign compression fractures of vertebrae on MRI. The authors performed a retrospective study including T1W1 and T2W1-FS images, as well as the T1W1/T2W1-FS combination, which served to validate a system called TSCCN, trained in a previous work. They compared the performance of the system with AI with respect to the perceptual response of expert human observers. Automatic 2D segmentations of the MRI containing the region of interest were used. Based on the area under the ROC curve, accuracy, and sensitivity and specificity metrics, the superiority of the system with AI in classification and its potential for future clinical applications was demonstrated. …