Article Figures & Data
Figures
- Fig 1.
Areas with lesion incidence are indicated with red boxes. A, Neuroradiologist manual segmentation of an axial section of the brain. B, Selected voxels for SuBLIME modeling. C, Axial section of the probability map from the full model. D, Axial section of the probability map from the SuBLIME model fit with only the T2-weighted image. E, Binary segmentation of the probability map from the full model with false-positive rate of 0.01 overlaid on the FLAIR image. F, Binary segmentation of the probability map from the full model with false-positive rate of 0.001 overlaid on the FLAIR image.
- Fig 2.
A, ROC curves for the voxel-level detection of incident and enlarging lesions for different thresholds of the probability maps produced from SuBLIME, as well as different thresholds of the T2-weighted subtraction without the model and voxel selection procedure. The red ROC curve is for the full model and has an AUC of 99% (95% CI [97%, 100%]). The blue ROC curve is for the model fit with only the T2-weighted image and has an AUC of 97% (95% CI [88%, 99%]). The green ROC curve is for thresholding the T2-weighted subtraction image without the model and voxel selection procedure and has an AUC of 92% (95% CI [83%, 95%]). B, Partial ROC curves for false-positive rates up to 0.01. The red curve is for the full model and the blue curve is for the model fit with only the T2-weighted image. The green curve is for thresholding the T2-weighted subtraction image without the model and voxel selection procedure.
Tables
Set Subtype Age Sex EDSS Treatment with Disease-Modifying Therapy (Baseline) Validation RRMS 37 Male 1.5 Yes Validation RRMS 38 Female 2.0 Yes Validation RRMS 48 Male 3.0 No Training RRMS 38 Female 1.5 No Training RRMS 30 Female 1.0 Yes Training RRMS 43 Female 1.5 Yes Validation RRMS 35 Female 1.5 Yes Training RRMS 37 Female 1.0 Yes Training RRMS 47 Female 3.0 Yes Validation RRMS 56 Female 1.0 No Note:—EDSS indicates Expanded Disability Status Scale; RRMS, relaping-remitting multiple sclerosis.
FA (degrees) TR (ms) TE (ms) TI (ms) FLAIR (90, 90) (10,000, 10,000) (77.8, 159.5) (2200, 2500) T2-weighted (90, 90) (3400, 6500) (94.6, 112.0) NA PD (90, 90) (3400, 6500) (11.8, 15.0) NA T1-weighted (13, 20) (7.68, 10.3) (1.88, 4.05) (450, 750)a Note:—FA indicates flip angle; NA, not available.
↵a 106 of the T1-weighted scans did not have an inversion time.
False- Positive Rate Threshold Value Specificity Sensitivity Volume Change (Actual 625 mm3) 0.01 0.0022 0.99 0.95 30454 0.001 0.0229 0.999 0.83 3520 0.00025 0.0815 0.99975 0.54 1082 0.0001 0.1396 0.9999 0.35 509
In this issue
Jump to section
Related Articles
Cited By...
- Evaluation of the Statistical Detection of Change Algorithm for Screening Patients with MS with New Lesion Activity on Longitudinal Brain MRI
- Evaluation of the Statistical Detection of Change Algorithm for Screening Patients with MS with New Lesion Activity on Longitudinal Brain MRI
- Evaluation of statistical detection of change algorithm for triaging multiple sclerosis patients with new lesion activity on longitudinal brain MRI
- Automatic brain lesion segmentation on standard magnetic resonance images: a scoping review
- Intensity warping for multisite MRI harmonization
- A New Approach to Symmetric Registration of Longitudinal Structural MRI of the Human Brain
- An automated statistical technique for counting distinct multiple sclerosis lesions can recover aspects of lesion history and provide relevant disease information
- Improved Detection of New MS Lesions during Follow-Up Using an Automated MR Coregistration-Fusion Method
- An Automated Statistical Technique for Counting Distinct Multiple Sclerosis Lesions
- Improved Automatic Detection of New T2 Lesions in Multiple Sclerosis Using Deformation Fields