NF1 MPNST Imaging Predictors (NTAP)

This Synapse project belongs to a Francis S. Collins Scholars Program study funded by the Neurofibromatosis Therapeutic Acceleration Program (NTAP) and led by Ina Ly at Massachusetts General Hospital. It gathers MRI scans and paired clinical data of people with neurofibromatosis type 1 (NF1) from several institutions. The aim is to find radiomic features (quantitative measures of tumor volume, intensity, shape and texture) that separate malignant peripheral nerve sheath tumors (MPNST) from the benign plexiform neurofibromas they usually arise from, and to build a model that flags patients at risk of malignant change. The NF Data Portal marks the study as completed and its data as under embargo.

Composition

The project page describes a plan, not a released cohort: MRI and clinical data from 200 NF1 patients at eight sites, split into a training set (60 MPNST and 60 plexiform neurofibroma), a validation set (20 and 20) and a test set (20 and 20). Those figures are targets and are not listed as counts here.

A 2023 American Academy of Neurology abstract by the study lead and co-authors reports 136 plexiform neurofibromas and 91 MPNSTs from nine centers. These are tumor counts, not patient counts, and the abstract does not state that this set is the one deposited on Synapse.

Acquisition

The abstract describes segmentation on STIR images where available, otherwise on T2-weighted or fat-saturated contrast-enhanced T1-weighted images, with masks registered to T1-weighted images. Images were bias-corrected, intensity-normalized and resampled to 1 mm voxels before feature extraction. Scanner details are not published. Which sequences are in the Synapse deposit is not stated, so no contrasts are listed in the entry.

Annotations

Tumors were segmented and labelled as plexiform neurofibroma or MPNST in the abstract's analysis. The project also holds folders for atypical neurofibromas and clinical data, which are not visible without access.

Known limitations

The data are under embargo. Access goes through the data contributor and no data use terms are published. Patient numbers, demographics and the content of the deposit cannot be verified from public sources.