Feasibility of Gadolinium Contrast Reduced Brain MRI: the Potential of Deep Learning

DeepMeds

Status and phase: Withdrawn; Phase 1

Conditions: Deep Learning; Artificial Intelligence

Treatments: Device: DeepGad

Study type: Interventional

Funder types: Industry

Identifiers

NCT06462924

DG53024

Details and patient eligibility

About

MRI scans were performed using 3 different 1.5T scanners with an eight-channel head coils. Following a 3D pre-contrast T1w scan, a low-dose contrast-enhanced 3D T1w scan was obtained using 20% (0.02 mmol/kg) of the standard dosage of gadoterate meglumine.

The subjects were immediately administered the remaining 80% (0.08 mmol/kg) of the contrast agent to reach the standard dose of 0.1 mmol/kg, which served as a training ground truth for further quantitative assessment. All three acquisitions were performed during a single imaging session, with no additional gadolinium dose administered above the standard protocol.

Sex: All

Ages: 18 to 80 years old

Volunteers: Accepts Healthy Volunteers

Inclusion criteria

1. Clinical indications for imaging with a contrast-enhanced 3D-T1w MRI sequence including tumor suspicion, postoperative tumor follow-up, multiple sclerosis, routine brain imaging, etc.,
2. no plan for dynamic contrast administration or deviation from the standard dose of 0.1 mmol/kg body weight (e.g., sella imaging, magnetic resonance angiography),
3. no clinical contraindications to imaging prolongation (i.e., emergency, poor patient condition).

Exclusion criteria

1. prominent image artifacts,
2. incomplete study sequences (e.g., early termination)
3. errors related to contrast agent administration.

Trial design

Primary purpose: Diagnostic

Allocation: N/A

Interventional model: Single Group Assignment

Masking: None (Open label)

---

0 participants in 1 patient group

DeepGad

Experimental group

Description:

The dataset comprises a total of 500 patients, with 300 patients used for model training and 200 patients reserved for model testing. Each patient had approximately 350 2D brain slices of the coregistered 3D volumes excluding the five slices at the base and five slices at the top of the acquired volume due to their low signal-to-noise ratio (SNR). For robustness and to avoid overfitting, standard DL data augmentation techniques consisting of eight types of transformations were applied to each case in the training set.

Treatment:

Device: DeepGad