Higher Validity, Lower Radiation: A New Ictal Single-Photon Emission Computed Tomography Framework.

OBJECTIVE: To assess whether arterial spin labeling perfusion images of healthy controls can enhance ictal single-photon emission computed tomography analysis and whether the acquisition of the interictal image can be omitted. METHODS: We developed 2 pipelines: The first uses ictal and interictal images and compares these to single-photon emission computed tomography and arterial spin labeling of healthy controls. The second pipeline uses only the ictal image and the analogous healthy controls. Both pipelines were compared to the gold standard analysis and evaluated on data of individuals with epilepsy who underwent ictal single-photon emission computed tomography imaging during presurgical evaluation between 2010 and 2022. Fifty healthy controls prospectively underwent arterial spin labeling imaging. The correspondence between the detected hyperperfusion and the postoperative resection cavity or the presumably affected lobe was assessed using Dice score and mean Euclidean distance. Additionally, the outcomes of the pipelines were automatically assigned to 1 of 5 concordance categories. RESULTS: Inclusion criteria were met by 43 individuals who underwent epilepsy surgery and by 73 non-surgical individuals with epilepsy. Compared to the gold standard analysis, both pipelines resulted in significantly higher Dice scores and lower mean distances (p < 0.05). The combination of both provided localizing results in 85/116 cases, compared to 54/116 generated by the current gold standard analysis and the ictal image alone produced localizing results in 60/116 (52%) cases. INTERPRETATION: We propose a new ictal single-photon emission computed tomography protocol; it finds relevantly more ictal hyperperfusion, and halves the radiation dose in about half of the individuals. ANN NEUROL 2024.

Artificial intelligence for the detection of focal cortical dysplasia: Challenges in translating algorithms into clinical practice.

Focal cortical dysplasias (FCDs) are malformations of cortical development and one of the most common pathologies causing pharmacoresistant focal epilepsy. Resective neurosurgery yields high success rates, especially if the full extent of the lesion is correctly identified and completely removed. The visual assessment of magnetic resonance imaging does not pinpoint the FCD in 30%-50% of cases, and half of all patients with FCD are not amenable to epilepsy surgery, partly because the FCD could not be sufficiently localized. Computational approaches to FCD detection are an active area of research, benefitting from advancements in computer vision. Automatic FCD detection is a significant challenge and one of the first clinical grounds where the application of artificial intelligence may translate into an advance for patients' health. The emergence of new methods from the combination of health and computer sciences creates novel challenges. Imaging data need to be organized into structured, well-annotated datasets and combined with other clinical information, such as histopathological subtypes or neuroimaging characteristics. Algorithmic output, that is, model prediction, requires a technically correct evaluation with adequate metrics that are understandable and usable for clinicians. Publication of code and data is necessary to make research accessible and reproducible. This critical review introduces the field of automatic FCD detection, explaining underlying medical and technical concepts, highlighting its challenges and current limitations, and providing a perspective for a novel research environment.

An open presurgery MRI dataset of people with epilepsy and focal cortical dysplasia type II.

Automated detection of lesions using artificial intelligence creates new standards in medical imaging. For people with epilepsy, automated detection of focal cortical dysplasias (FCDs) is widely used because subtle FCDs often escape conventional neuroradiological diagnosis. Accurate recognition of FCDs, however, is of outstanding importance for affected people, as surgical resection of the dysplastic cortex is associated with a high chance of postsurgical seizure freedom. Here, we make publicly available a dataset of 85 people affected by epilepsy due to FCD type II and 85 healthy control persons. We publish 3D-T1 and 3D-FLAIR, manually labeled regions of interest, and carefully selected clinical features. The open presurgery MRI dataset may be used to validate existing automated algorithms of FCD detection as well as to create new approaches. Most importantly, it will enable comparability of already existing approaches and support a more widespread use of automated lesion detection tools.