Proposed reporting items for automatic brain lesion segmentation studies
| Technical validation | Method | List of processing steps necessary to apply to a raw image |
| Computational system parameters | ||
| Computation time | ||
| Open documentation of the algorithm | ||
| Reference segmentation | Number of raters | |
| Raters’ training/experience | ||
| Method of segmentation | ||
| Method of consolidation (if multiple raters) | ||
| Validation | List of validation metrics used | |
| Number of images (split into training/validation/testing if applicable) | ||
| Input sequences | ||
| Number of scanners used to acquire images | ||
| Acquisition parameters | ||
| Number of time points and intervals for longitudinal data | ||
| Results | Mean, median, SD for each validation metric | |
| Number of failed cases (if applicable) | ||
| Preclinical validation | Patient information | Diagnosis and level of verification for example, clinical follow-up, tissue sampling, autopsy, etc. |
| Clinical presentation | ||
| Administered treatments (if applicable) | ||
| Clinical task | Explicit definition of the clinical task for which the algorithm is applied (eg, lesion growth estimation, treatment evaluation, radiotherapy planning). | |
| User validation | List of optimisation metrics used for the clinical task (if applicable). | |
| User’s method of evaluating the outcome. | ||
| Clinical validation | Information and storage system compatibility | Compatibility with picture archiving, communication, and storage systems |
| Regulatory approval | Compliance with the regulatory approvals for software as a medical device. |