Comparative Study of Conventional MR Images With Synthetically Reconstructed MR Images of the Brain

Sensitivity of pathological findings (true number of participants with pathological findings over assessed number of participants with pathological findings by reader) for synthetic and conventional MR images on the full analysis set. Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Specificity of pathological findings (true number of participants with normal/no finding over assessed number of participants with normal/no finding by reader) for synthetic and conventional MR images on the full analysis set. Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Diagnostic accuracy of radiological findings (radiological finding class) for synthetic and conventional MR images on the full analysis set. Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Sensitivity of pathological findings (true number of participants with pathological findings over assessed number of participants with pathological findings) for synthetic and conventional MR images in adult population (18-99 years). Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Specificity of pathological findings (true number of participants with normal/no finding over assessed number of participants with normal/no finding) for synthetic and conventional MR images on the in adult population (18-99 years). Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Diagnostic accuracy of radiological findings (radiological finding class) for synthetic and conventional MR images in adult population (18-99 years). Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Sensitivity of pathological findings (true number of participants with pathological findings over assessed number of participants with pathological findings) for synthetic and conventional MR images in pediatric population (0-17 years). Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Specificity of pathological findings (true number of participants with normal/no finding over assessed number of participants with normal/no finding) for synthetic and conventional MR images on the in pediatric population (0-17 years). Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Diagnostic accuracy of radiological findings (radiological finding class) for synthetic and conventional MR images in pediatric population (0-17 years). Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The ground truth is the site-determined diagnosis.

Legibility of anatomical structures, per structure, on the full analysis set for T1w images. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The result for each anatomical structure per image type, independent of reading period, is summarized in corresponding row below.

Legibility of anatomical structures, per structure, on the full analysis set for T2w images. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The result for each anatomical structure per image type, independent of reading period, is summarized in corresponding row below.

Presence and type of artifacts determined for synthetic T1w images, on the full analysis set determined by all readers. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The result for each artifact per image type, independent of reading period, is summarized in corresponding row below.

Presence and type of artifacts determined for synthetic T2w images, on the full analysis set determined by all readers. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The result for each artifact per image type, independent of reading period, is summarized in corresponding row below.

Inter-rater agreement between all reader's assessment of pathological finding, as well as radiological finding classes, on the full analysis set. Each image was evaluated only once per reader and the agreement is independent of reading session. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The agreement was assessed separately per image type over both reading periods using overall agreement, Fleiss' kappa coefficient, and Gwet's AC1 coefficient. The first reading begun after all image processing was complete.

Inter-method agreement between the different reader's assessment of pathological finding, as well as radiological finding classes, on the full analysis set. Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The agreement was evaluated for all images, independent of reading period and image type, using overall agreement, Fleiss' kappa coefficient, and Gwet's AC1 coefficient. The first reading begun after all image processing was complete.

Image quality score (1 (Unacceptable) - 5 (Excellent)) per contrast weighting (T1w, T2w) per both image types (conventional and synthetic) where 3,4,5 are deemed acceptable for diagnostic use and 1,2 unacceptable for diagnostic use. Each image was evaluated only once per reader. The images were split between two readings, with a wash-out period of 4 weeks in between. Both readings contained both types of images, synthetic and conventional. The first reading begun after all image processing was complete. The result presented, independent of reading period, is the averaged score per contrast weight and image type with the standard deviation.

Incidence of Adverse Events (AE), Adverse Device Effects (ADE), Serious Adverse Events (SAE), Serious Adverse Device Effects (SADE), Unanticipated Adverse Device Effect (UADE), Unanticipated Serious Adverse Device Effect (USADE) and Device Deficiencies (DD). The measurement was assessed in two periods: during exam day after signing consent form (all incidents applicable) and when generating the synthetic images (only device deficiencies) after all exams were performed.