For clinicians managing medically refractory focal epilepsy, a major challenge often lies in what are referred to as “nonlesional” brain MRI scans. Subtle focal cortical dysplasia (FCD) often cannot be visually assessed with standard MRI, resulting in epilepsy surgery being canceled or postponed, or resulting in unsuccessful surgical outcomes.

Two recent studies from the Cleveland Clinic demonstrate that magnetic resonance fingerprinting (MRF), a rapid quantitative imaging technique, may significantly improve the ability to detect elusive FCD lesions and determine which lesions are actively causing a patient’s seizures.

These studies collectively demonstrate progress in defining the optimal clinical applicability of MRF when conventional MRI is inadequate in epilepsy imaging. “We have shown that by integrating MRF with machine learning and surface-based analysis, clinicians can achieve high detection sensitivity of subtle FCD while simultaneously reducing the noise of false positives,” said the study’s senior author. Dr. Eileen WangDirector of Research, Cleveland Clinic Epilepsy Center. “We also found that in complex cases with multiple malformations, MRF provides a non-invasive method to prioritize which lesions require invasive electroencephalography (SEEG) exploration.”

Research 1: Automatic detection and subtyping using machine learning

First study (epilepsy. 2025 Epub 9 October), led by Dr. Wang and Dr. Ting-Yu Su, a postdoctoral fellow in her lab, focused on developing a robust framework for whole-brain FCD detection by combining MRF with machine learning and surface-based morphometry, a widely adopted MRI post-processing technique.

“Manual identification of FCD is highly variable and dependent on specialized clinical expertise,” explains Dr. Su. “Subtle image features are easy to miss. We aimed to create an automated pipeline that leverages MRF’s high-resolution quantitative data to improve existing surface-based detection methods.”

Study design and results

The research team retrospectively analyzed 44 patients with confirmed FCD who underwent MRF research scans at the Cleveland Clinic. They also recruited 70 age- and gender-matched healthy individuals to undergo MRF scanning as a control group.

All MRF imaging was performed using high-resolution 3T MRF sequences (scan time approximately 10 min) to generate T1 and T2 maps. These maps were integrated with clinically acquired structural T1-weighted and 3D FLAIR images. Quantitative features were generated based on all these data.

For FCD detection, the researchers adopted a two-step machine learning approach using:

• Vertex-oriented neural network classifier for identifying potentially abnormal cortical vertices.
• Per-cluster random undersampling boosting classifier specifically designed to prevent false positives by analyzing cluster size and feature statistics

Key findings include:

• Improved sensitivity. Combining T1-weighted images with MRF and FLAIR resulted in a sensitivity of 71.4% for FCD detection. In contrast, clinical MRI detected only 57% of lesions based on official radiology reports.
• Better control of false positives. The addition of MRF data significantly reduced false-positive clusters in patients and controls, in some cases by more than 50%.
• Accurate subtyping. This framework distinguished type II FCD malformations from non-type II malformations with 80.8% accuracy and outperformed traditional radiological markers such as transmembrane signs in classifying the FCD IIb subtype.
• Interpretability. Across all feature sets, true positive clusters had consistently higher detection probabilities than false positive clusters. The researchers noted that this separation suggests that the probability value may serve as a practical confidence measure, strengthening the rationale for clinical adoption.
• Correlation of results. This framework showed higher sensitivity for FCD detection in patients who were seizure-free after surgery (72.7%) compared to those who were not (50.0%), suggesting that its output may serve as an indicator of seizure outcome.

what it means

“This study is the first to integrate surface-based morphometry with MRF-generated quantitative tissue property maps,” said Dr. Wang. “This proves that an MRF-driven machine learning framework can match or exceed the performance of human experts in lesion detection. For clinicians, the framework’s predicted probabilities can serve as a confidence metric in distinguishing true lesions from anatomical variations and image artifacts.”

Study 2: Distinguishing between epileptogenic and silent malformations

The second study (J Neurol Sci. 2025;477:12351) addressed another clinical dilemma: in patients with multiple cortical malformations, which is the cause of seizures?

Although multiple FCD-like abnormalities or extensive polymicrogyria (PMGs) are present in a single patient, these lesions are not always equally epileptogenic. Determining which lesions to target often requires invasive stereoelectroencephalography (SEEG). In this pilot study, we investigated whether MRF could noninvasively indicate the active seizure onset zone in this setting.

Study design and results

Dr. Wang and colleagues retrospectively analyzed 69 people who underwent 3D whole-brain MRF research scans at the Cleveland Clinic. This included 21 patients with refractory focal epilepsy and 48 healthy controls who were included for comparative analysis. Of the patients, 4 had complex cortical malformations (FCD or PMG) and underwent SEEG and/or surgery, and 17 had histopathologically confirmed FCD II.

Among the four complex cases, the researchers compared MRF signatures within the same patient (comparing active lesions to silent lesions) and between patients (comparing lesions to those of 17 patients with FCD II).

Key results include:

• Gray matter T1 as a signature of seizures. In all four complex cases, epileptogenic malformations showed significantly higher gray matter T1 values ​​compared to non-epileptogenic areas. This finding holds both within and between individual patients.
• Ability to reveal MRI-negative lesions. In one complicated case, a lesion was identified by conventional MRI but was ultimately found to be non-epileptogenic. Conversely, MRF detected elevated T1 and T2 values ​​in another MRI-negative region, which was later confirmed by SEEG to be the true seizure onset region.

what it means

Dr. Wang and his coauthors conclude that quantitative gray matter T1 measured by MRF may serve as a sensitive marker of epileptogenicity. “MRF holds promise as a non-invasive imaging probe for in vivo epileptogenicity,” says Dr. Wang. “What we need to further understand is which quantitative indicators are consistently and strongly correlated with epileptogenic tissue. If our preliminary findings are confirmed in larger studies, MRF may play a role in guiding SEEG implantation and surgical planning in complex cases involving multiple cortical malformations.”

FCD MRF Key Points

Together, these two studies provide some insight into the management of refractory epilepsy, according to researchers at the Cleveland Clinic Epilepsy Center.

• Value beyond visual detection. Clinicians should view MRF not only as a tool for detecting lesions, but also as a way to characterize tissue pathology. The ability to provide absolute values ​​allows comparison of a patient’s brain to a standard pathology library, aiding in the diagnosis of MRI-negative cases.
• Role in optimizing SEEG transplantation. When multiple lesions are present, elevated T1 gray matter values ​​on the MRF can help guide electrode placement. Lesions that appear abnormal on structural scans but have normal MRF values ​​may be less likely to be the primary seizure source.
• Potential to reduce false positives in automatic detection. One hurdle to AI-based lesion detection is high false-positive rates, which can burden clinical reviews. These studies demonstrate that the quantitative nature of MRF may be invaluable in filtering artifacts that appear to be FCD but lack underlying tissue property changes associated with the disease.
• Prognostic value. The correlation between model detection and seizure-free outcome suggests that MRF-based measurements may provide valuable non-invasive biomarkers of outcome.

“As MRF moves toward broader multicenter validation, the fact that it can now be acquired in just a few minutes makes it increasingly practical for routine use,” concludes Dr. Wang. “It could be an important part of preoperative evaluation and surgical planning for patients with refractory focal epilepsy.”