ABSTRACT
OBJECTIVES: To develop a multiparametric machine-learning (ML) framework using high-resolution 3 dimensional (3D) magnetic resonance (MR) fingerprinting (MRF) data for quantitative characterization of focal cortical dysplasia (FCD). MATERIALS: We included 119 subjects, 33 patients with focal epilepsy and histopathologically confirmed FCD, 60 age- and gender-matched healthy controls (HCs), and 26 disease controls (DCs). Subjects underwent whole-brain 3 Tesla MRF acquisition, the reconstruction of which generated T1 and T2 relaxometry maps. A 3D region of interest was manually created for each lesion, and z-score normalization using HC data was performed. We conducted 2D classification with ensemble models using MRF T1 and T2 mean and standard deviation from gray matter and white matter for FCD versus controls. Subtype classification additionally incorporated entropy and uniformity of MRF metrics, as well as morphometric features from the morphometric analysis program (MAP). We translated 2D results to individual probabilities using the percentage of slices above an adaptive threshold. These probabilities and clinical variables were input into a support vector machine for individual-level classification. Fivefold cross-validation was performed and performance metrics were reported using receiver-operating-characteristic-curve analyses. RESULTS: FCD versus HC classification yielded mean sensitivity, specificity, and accuracy of 0.945, 0.980, and 0.962, respectively; FCD versus DC classification achieved 0.918, 0.965, and 0.939. In comparison, visual review of the clinical magnetic resonance imaging (MRI) detected 48% (16/33) of the lesions by official radiology report. In the subgroup where both clinical MRI and MAP were negative, the MRF-ML models correctly distinguished FCD patients from HCs and DCs in 98.3% of cross-validation trials for the magnetic resonance imaging negative group and MAP negative group. Type II versus non-type-II classification exhibited mean sensitivity, specificity, and accuracy of 0.835, 0.823, and 0.83, respectively; type IIa versus IIb classification showed 0.85, 0.9, and 0.87. In comparison, the transmantle sign was present in 58% (7/12) of the IIb cases. INTERPRETATION: The MRF-ML framework presented in this study demonstrated strong efficacy in noninvasively classifying FCD from normal cortex and distinguishing FCD subtypes. ANN NEUROL 2024.
ABSTRACT
We developed a scoring system for assessing glaucoma risk using demographic and laboratory factors by employing a no-code approach (automated coding) using ChatGPT-4. Comprehensive health checkup data were collected from the Korea National Health and Nutrition Examination Survey. Using ChatGPT-4, logistic regression was conducted to predict glaucoma without coding or manual numerical processes, and the scoring system was developed based on the odds ratios (ORs). ChatGPT-4 also facilitated the no-code creation of an easy-to-use risk calculator for glaucoma. The ORs for the high-risk groups were calculated to measure performance. ChatGPT-4 automatically developed a scoring system based on demographic and laboratory factors, and successfully implemented a risk calculator tool. The predictive ability of the scoring system was comparable to that of traditional machine learning approaches. For high-risk groups with 1-2, 3-4, and 5 + points, the calculated ORs for glaucoma were 1.87, 2.72, and 15.36 in the validation set, respectively, compared with the group with 0 or fewer points. This study presented a novel no-code approach for developing a glaucoma risk assessment tool using ChatGPT-4, highlighting its potential for democratizing advanced predictive analytics, making them readily available for clinical use in glaucoma detection.
ABSTRACT
OBJECTIVE: We aim to improve focal cortical dysplasia (FCD) detection by combining high-resolution, three-dimensional (3D) magnetic resonance fingerprinting (MRF) with voxel-based morphometric magnetic resonance imaging (MRI) analysis. METHODS: We included 37 patients with pharmacoresistant focal epilepsy and FCD (10 IIa, 15 IIb, 10 mild Malformation of Cortical Development [mMCD], and 2 mMCD with oligodendroglial hyperplasia and epilepsy [MOGHE]). Fifty-nine healthy controls (HCs) were also included. 3D lesion labels were manually created. Whole-brain MRF scans were obtained with 1 mm3 isotropic resolution, from which quantitative T1 and T2 maps were reconstructed. Voxel-based MRI postprocessing, implemented with the morphometric analysis program (MAP18), was performed for FCD detection using clinical T1w images, outputting clusters with voxel-wise lesion probabilities. Average MRF T1 and T2 were calculated in each cluster from MAP18 output for gray matter (GM) and white matter (WM) separately. Normalized MRF T1 and T2 were calculated by z-scores using HCs. Clusters that overlapped with the lesion labels were considered true positives (TPs); clusters with no overlap were considered false positives (FPs). Two-sample t-tests were performed to compare MRF measures between TP/FP clusters. A neural network model was trained using MRF values and cluster volume to distinguish TP/FP clusters. Ten-fold cross-validation was used to evaluate model performance at the cluster level. Leave-one-patient-out cross-validation was used to evaluate performance at the patient level. RESULTS: MRF metrics were significantly higher in TP than FP clusters, including GM T1, normalized WM T1, and normalized WM T2. The neural network model with normalized MRF measures and cluster volume as input achieved mean area under the curve (AUC) of .83, sensitivity of 82.1%, and specificity of 71.7%. This model showed superior performance over direct thresholding of MAP18 FCD probability map at both the cluster and patient levels, eliminating ≥75% FP clusters in 30% of patients and ≥50% of FP clusters in 91% of patients. SIGNIFICANCE: This pilot study suggests the efficacy of MRF for reducing FPs in FCD detection, due to its quantitative values reflecting in vivo pathological changes. © 2024 International League Against Epilepsy.
Subject(s)
Magnetic Resonance Imaging , Malformations of Cortical Development , Humans , Magnetic Resonance Imaging/methods , Female , Male , Adult , Malformations of Cortical Development/diagnostic imaging , Malformations of Cortical Development/pathology , Adolescent , Young Adult , Epilepsies, Partial/diagnostic imaging , Epilepsies, Partial/pathology , Middle Aged , Drug Resistant Epilepsy/diagnostic imaging , Drug Resistant Epilepsy/pathology , Imaging, Three-Dimensional/methods , Child , False Positive Reactions , Gray Matter/diagnostic imaging , Gray Matter/pathology , Image Processing, Computer-Assisted/methods , Focal Cortical DysplasiaABSTRACT
BACKGROUND: The epiretinal membrane (ERM) is a common retinal disorder characterized by abnormal fibrocellular tissue at the vitreomacular interface. Most patients with ERM are asymptomatic at early stages. Therefore, screening for ERM will become increasingly important. Despite the high prevalence of ERM, few deep learning studies have investigated ERM detection in the color fundus photography (CFP) domain. In this study, we built a generative model to enhance ERM detection performance in the CFP. METHODS: This deep learning study retrospectively collected 302 ERM and 1,250 healthy CFP data points from a healthcare center. The generative model using StyleGAN2 was trained using single-center data. EfficientNetB0 with StyleGAN2-based augmentation was validated using independent internal single-center data and external datasets. We randomly assigned healthcare center data to the development (80%) and internal validation (20%) datasets. Data from two publicly accessible sources were used as external validation datasets. RESULTS: StyleGAN2 facilitated realistic CFP synthesis with the characteristic cellophane reflex features of the ERM. The proposed method with StyleGAN2-based augmentation outperformed the typical transfer learning without a generative adversarial network. The proposed model achieved an area under the receiver operating characteristic (AUC) curve of 0.926 for internal validation. AUCs of 0.951 and 0.914 were obtained for the two external validation datasets. Compared with the deep learning model without augmentation, StyleGAN2-based augmentation improved the detection performance and contributed to the focus on the location of the ERM. CONCLUSIONS: We proposed an ERM detection model by synthesizing realistic CFP images with the pathological features of ERM through generative deep learning. We believe that our deep learning framework will help achieve a more accurate detection of ERM in a limited data setting.
Subject(s)
Deep Learning , Epiretinal Membrane , Humans , Epiretinal Membrane/diagnostic imaging , Retrospective Studies , Diagnostic Techniques, Ophthalmological , Photography/methodsABSTRACT
Recently, fundus photography (FP) is being increasingly used. Corneal curvature is an essential factor in refractive errors and is associated with several pathological corneal conditions. As FP-based examination systems have already been widely distributed, it would be helpful for telemedicine to extract information such as corneal curvature using FP. This study aims to develop a deep learning model based on FP for corneal curvature prediction by categorizing corneas into steep, regular, and flat groups. The EfficientNetB0 architecture with transfer learning was used to learn FP patterns to predict flat, regular, and steep corneas. In validation, the model achieved a multiclass accuracy of 0.727, a Matthews correlation coefficient of 0.519, and an unweighted Cohen's κ of 0.590. The areas under the receiver operating characteristic curves for binary prediction of flat and steep corneas were 0.863 and 0.848, respectively. The optic nerve and its peripheral areas were the main focus of the model. The developed algorithm shows that FP can potentially be used as an imaging modality to estimate corneal curvature in the post-COVID-19 era, whereby patients may benefit from the detection of abnormal corneal curvatures using FP in the telemedicine setting.
Subject(s)
COVID-19 , Deep Learning , Humans , Diagnostic Techniques, Ophthalmological , Cornea/diagnostic imaging , PhotographyABSTRACT
Traumatic brain injury (TBI) triggers progressive neurodegeneration resulting in brain atrophy that continues months-to-years following injury. However, a comprehensive characterization of the spatial and temporal evolution of TBI-related brain atrophy remains incomplete. Utilizing a sensitive and unbiased morphometry analysis pipeline optimized for detecting longitudinal changes, we analyzed a sample consisting of 37 individuals with moderate-severe TBI who had primarily high-velocity and high-impact injury mechanisms. They were scanned up to three times during the first year after injury (3 months, 6 months, and 12 months post-injury) and compared with 33 demographically matched controls who were scanned once. Individuals with TBI already showed cortical thinning in frontal and temporal regions and reduced volume in the bilateral thalami at 3 months post-injury. Longitudinally, only a subset of cortical regions in the parietal and occipital lobes showed continued atrophy from 3 to 12 months post-injury. Additionally, cortical white matter volume and nearly all deep gray matter structures exhibited progressive atrophy over this period. Finally, we found that disproportionate atrophy of cortex along sulci relative to gyri, an emerging morphometric marker of chronic TBI, was present as early as 3 month post-injury. In parallel, neurocognitive functioning largely recovered during this period despite this pervasive atrophy. Our findings demonstrate msTBI results in characteristic progressive neurodegeneration patterns that are divergent across regions and scale with the severity of injury. Future clinical research using atrophy during the first year of TBI as a biomarker of neurodegeneration should consider the spatiotemporal profile of atrophy described in this study.
Subject(s)
Brain Injuries, Traumatic , Brain Injuries , Brain Injury, Chronic , White Matter , Humans , Brain Injuries, Traumatic/complications , Brain Injuries, Traumatic/diagnostic imaging , Brain Injuries, Traumatic/pathology , Brain Injuries/pathology , White Matter/pathology , Atrophy/pathology , Brain/diagnostic imaging , Brain/pathologyABSTRACT
OBJECTIVE: We aim to quantify whole-brain tissue-property changes in patients with magnetic resonance imaging (MRI)-negative pharmacoresistant focal epilepsy by three-dimensional (3D) magnetic resonance fingerprinting (MRF). METHODS: We included 30 patients with pharmacoresistant focal epilepsy and negative MRI by official radiology report, as well as 40 age- and gender-matched healthy controls (HCs). MRF scans were obtained with 1 mm3 isotropic resolution. Quantitative T1 and T2 relaxometry maps were reconstructed from MRF and registered to the Montreal Neurological Institute (MNI) space. A two-sample t test was performed in Functional Magnetic Resonance Imaging of the Brain (FMRIB) Software Library (FSL) to evaluate significant abnormalities in patients comparing to HCs, with correction by the threshold-free cluster enhancement (TFCE) method. Subgroups analyses were performed for extra-temporal epilepsy/temporal epilepsy (ETLE/TLE), and for those with/without subtle abnormalities detected by morphometric analysis program (MAP), to investigate each subgroup's pattern of MRF changes. Correlation analyses were performed between the mean MRF values in each significant cluster and seizure-related clinical variables. RESULTS: Compared to HCs, patients exhibited significant group-level T1 increase ipsilateral to the epileptic origin, in the mesial temporal gray matter (GM) and white matter (WM), temporal pole GM, orbitofrontal GM, hippocampus, and amygdala, with scattered clusters in the neocortical temporal and insular GM. No significant T2 changes were detected. The ETLE subgroup showed a T1-increase pattern similar to the overall cohort, with additional involvement of the ipsilateral anterior cingulate GM. The subgroup of MAP+ patients also showed a T1-increase pattern similar to the overall cohort, with additional cluster in the ipsilateral lateral orbitofrontal GM. Higher T1 was associated with younger seizure-onset age, longer epilepsy duration, and higher seizure frequency. SIGNIFICANCE: MRF revealed group-level T1 increase in limbic/paralimbic structures ipsilateral to the epileptic origin, in patients with pharmacoresistant focal epilepsy and no apparent lesions on MRI, suggesting that these regions may be commonly affected by seizures in the epileptic brain. The significant association between T1 increase and higher seizure burden may reflect progressive tissue damage.
Subject(s)
Epilepsies, Partial , Epilepsy , Humans , Brain/diagnostic imaging , Magnetic Resonance Imaging/methods , Seizures , Epilepsies, Partial/diagnostic imagingABSTRACT
Quantitative magnetic resonance (MR) has been used to study cyto- and myelo-architecture of the human brain non-invasively. However, analyzing brain cortex using high-resolution quantitative MR acquisition can be challenging to perform using 3T clinical scanners. MR fingerprinting (MRF) is a highly efficient and clinically feasible quantitative MR technique that simultaneously provides T1 and T2 relaxation maps. Using 3D MRF from 40 healthy subjects (mean age = 25.6 ± 4.3 years) scanned on 3T magnetic resonance imaging, we generated whole-brain gyral-based normative MR relaxation atlases and investigated cortical-region-based T1 and T2 variations. Gender and age dependency of T1 and T2 variations were additionally analyzed. The coefficient of variation of T1 and T2 for each cortical-region was 3.5% and 7.3%, respectively, supporting low variability of MRF measurements across subjects. Significant differences in T1 and T2 were identified among 34 brain regions (P < 0.001), lower in the precentral, postcentral, paracentral lobule, transverse temporal, lateral occipital, and cingulate areas, which contain sensorimotor, auditory, visual, and limbic functions. Significant correlations were identified between age and T1 and T2 values. This study established whole-brain MRF T1 and T2 atlases of healthy subjects using a clinical 3T scanner, which can provide a quantitative and region-specific baseline for future brain studies and pathology detection.
Subject(s)
Brain , Magnetic Resonance Imaging , Humans , Young Adult , Adult , Infant , Magnetic Resonance Imaging/methods , Magnetic Resonance Spectroscopy , Phantoms, Imaging , Healthy Volunteers , Image Processing, Computer-Assisted/methodsABSTRACT
BACKGROUND AND OBJECTIVES: We aim to provide detailed imaging-electroclinicopathologic characterization of the black line sign, a novel MRI marker for focal cortical dysplasia (FCD) IIB. METHODS: 7T T2*-weighted gradient-echo (T2*w-GRE) images were retrospectively reviewed in a consecutive cohort of patients with medically intractable epilepsy with pathology-proven FCD II, for the occurrence of the black line sign. We examined the overlap between the black line region and the seizure-onset zone (SOZ) defined by intracranial EEG (ICEEG) and additionally assessed whether complete inclusion of the black line region in the surgical resection was associated with postoperative seizure freedom. The histopathologic specimen was aligned with the MRI to investigate the pathologic underpinning of the black line sign. Region-of-interest-based quantitative MRI (qMRI) analysis on the 7T T1 map was performed in the black line region, entire lesional gray matter (GM), and contralateral/ipsilateral normal gray and white matter (WM). RESULTS: We included 20 patients with FCD II (14 IIB and 6 IIA). The black line sign was identified in 12/14 (85.7%) of FCD IIB and 0/6 of FCD IIA on 7T T2*w-GRE. The black line region was highly concordant with the ICEEG-defined SOZ (5/7 complete and 2/7 partial overlap). Seizure freedom was seen in 8/8 patients whose black line region was completely included in the surgical resection; in the 2 patients whose resection did not completely include the black line region, both had recurring seizures. Inclusion of the black line region in the surgical resection was significantly associated with seizure freedom (p = 0.02). QMRI analyses showed that the T1 mean value of the black line region was significantly different from the WM (p < 0.001), but similar to the GM. Well-matched histopathologic slices in one case revealed accumulated dysmorphic neurons and balloon cells in the black line region. DISCUSSION: The black line sign may serve as a noninvasive marker for FCD IIB. Both MRI-pathology and qMRI analyses suggest that the black line region was an abnormal GM component within the FCD. Being highly concordant with ICEEG-defined SOZ and significantly associated with seizure freedom when included in resection, the black line sign may contribute to the planning of ICEEG/surgery of patients with medically intractable epilepsy with FCD IIB. CLASSIFICATION OF EVIDENCE: This study provides Class II evidence that in individuals with intractable focal epilepsy undergoing resection who have a 7T MRI with adequate image quality, the presence of the black line sign may suggest FCD IIB, be concordant with SOZ from ICEEG, and be associated with more seizure freedom if fully included in resection.
Subject(s)
Drug Resistant Epilepsy , Epilepsies, Partial , Malformations of Cortical Development , Drug Resistant Epilepsy/complications , Drug Resistant Epilepsy/diagnostic imaging , Drug Resistant Epilepsy/surgery , Epilepsies, Partial/complications , Humans , Magnetic Resonance Imaging/methods , Malformations of Cortical Development/complications , Malformations of Cortical Development/diagnostic imaging , Malformations of Cortical Development/surgery , Retrospective Studies , Seizures/complicationsABSTRACT
OBJECTIVES: Magnetic resonance fingerprinting (MRF) is a novel, quantitative, and noninvasive technique to measure brain tissue properties. We aim to use MRF for characterizing normal-appearing thalamic and basal ganglia nuclei in the epileptic brain. METHODS: A three-dimensional (3D) MRF protocol (1 mm3 isotropic resolution) was acquired from 48 patients with unilateral medically intractable focal epilepsy and 39 healthy controls (HCs). Whole-brain T1 and T2 maps (containing T1 and T2 relaxation times) were reconstructed for each subject. Ten subcortical nuclei in the thalamus and basal ganglia were segmented as regions of interest (ROIs), within which the mean T1 and T2 values, as well as their coefficient of variation (CV) were compared between the patients and HCs at the group level. Subgroup and correlation analyses were performed to examine the relationship between significant MRF measures and various clinical characteristics. Using significantly abnormal MRF measures from the group-level analyses, support vector machine (SVM) and logistic regression machine learning models were built and tested with 5-fold and 10-fold cross-validations, to separate patients from HCs, and to separate patients with left-sided and right-sided epilepsy, at the individual level. RESULTS: MRF revealed increased T1 mean value in the ipsilateral thalamus and nucleus accumbens; increased T1 CV in the bilateral thalamus, bilateral pallidum, and ipsilateral caudate; and increased T2 CV in the ipsilateral thalamus in patients compared to HCs (p < .05, false discovery rate [FDR] corrected). The SVM classifier produced 78.2% average accuracy to separate individual patients from HCs, with an area under the curve (AUC) of 0.83. The logistic regression classifier produced 67.4% average accuracy to separate patients with left-sided and right-sided epilepsy, with an AUC of 0.72. SIGNIFICANCE: MRF revealed bilateral tissue-property changes in the normal-appearing thalamus and basal ganglia, with ipsilateral predominance and thalamic preference, suggesting subcortical involvement/impairment in patients with medically intractable focal epilepsy. The individual-level performance of the MRF-based machine-learning models suggests potential opportunities for predicting lateralization.
Subject(s)
Drug Resistant Epilepsy , Epilepsies, Partial , Epilepsy , Basal Ganglia/diagnostic imaging , Drug Resistant Epilepsy/diagnostic imaging , Epilepsies, Partial/diagnostic imaging , Humans , Magnetic Resonance Imaging/methods , Thalamus/diagnostic imagingABSTRACT
OBJECTIVE: We aimed to use a novel magnetic resonance fingerprinting (MRF) technique to examine in vivo tissue property characteristics of periventricular nodular heterotopia (PVNH). These characteristics were further correlated with stereotactic-electroencephalographic (SEEG) ictal onset findings. METHODS: We included five patients with PVNH who had SEEG-guided surgery and at least 1 year of seizure freedom or substantial seizure reduction. High-resolution MRF scans were acquired at 3 T, generating three-dimensional quantitative T1 and T2 maps. We assessed the differences between T1 and T2 values from the voxels in the nodules located in the SEEG-defined seizure onset zone (SOZ) and non-SOZ, on -individual and group levels. Receiver operating characteristic analyses were performed to obtain the optimal classification performance. Quantification of SEEG ictal onset signals from the nodules was performed by calculating power spectrum density (PSD). The association between PSD and T1 /T2 values was further assessed at different frequency bands. RESULTS: Individual-level analysis showed T1 was significantly higher in SOZ voxels than non-SOZ voxels (p < .05), with an average 73% classification accuracy. Group-level analysis also showed higher T1 was significantly associated with SOZ voxels (p < .001). At the optimal cutoff (normalized T1 of 1.1), a 76% accuracy for classifying SOZ nodules from non-SOZ nodules was achieved. T1 values were significantly associated with ictal onset PSD at the ultraslow, θ, ß, γ, and ripple bands (p < .05). T2 values were significantly associated with PSD only at the ultraslow band (p < .05). SIGNIFICANCE: Quantitative MRF measures, especially T1 , can provide additional noninvasive information to separate nodules in SOZ and non-SOZ. The T1 and T2 tissue property changes carry electrophysiological underpinnings relevant to the epilepsy, as shown by their significant positive associations with power changes during the SEEG seizure onset. The use of MRF as a supplementary noninvasive tool may improve presurgical evaluation for patients with PVNH and pharmacoresistant epilepsy.
Subject(s)
Epilepsy , Periventricular Nodular Heterotopia , Electroencephalography/methods , Humans , Magnetic Resonance Imaging/methods , Magnetic Resonance Spectroscopy , Periventricular Nodular Heterotopia/complications , Seizures/complicationsABSTRACT
OBJECTIVE: To examine the individual-patient-level localization value of resting-state functional MRI (rsfMRI) metrics for the seizure onset zone (SOZ) defined by stereo-electroencephalography (SEEG) in patients with medically intractable focal epilepsies. METHODS: We retrospectively included 19 patients who underwent SEEG implantation for epilepsy presurgical evaluation. Voxel-wise whole-brain analysis was performed on 3.0 T rsfMRI to generate clusters for amplitude of low-frequency fluctuations (ALFF), regional homogeneity (ReHo) and degree centrality (DC), which were co-registered with the SEEG-defined SOZ to evaluate their spatial overlap. Subgroup and correlation analyses were conducted for various clinical characteristics. RESULTS: ALFF demonstrated concordant clusters with SEEG-defined SOZ in 73.7% of patients, with 93.3% sensitivity and 77.8% PPV. The concordance rate showed no significant difference when subgrouped by lesional/non-lesional MRI, SOZ location, interictal epileptiform discharges on scalp EEG, pathology or seizure outcomes. No significant correlation was seen between ALFF concordance rate and epilepsy duration, seizure-onset age, seizure frequency or number of antiseizure medications. ReHo and DC did not achieve favorable concordance results (10.5% and 15.8%, respectively). All concordant clusters showed regional activation, representing increased neural activities. CONCLUSION: ALFF had high concordance rate with SEEG-defined SOZ at individual-patient level. SIGNIFICANCE: ALFF activation on rsfMRI can add localizing information for the noninvasive presurgical workup of intractable focal epilepsies.
Subject(s)
Anterior Temporal Lobectomy/methods , Electroencephalography/methods , Epilepsy/diagnostic imaging , Magnetic Resonance Imaging/methods , Adolescent , Adult , Anterior Temporal Lobectomy/adverse effects , Epilepsy/physiopathology , Epilepsy/surgery , Female , Humans , Male , Middle Aged , Postoperative Complications/epidemiology , Preoperative PeriodABSTRACT
Multimodal image integration (MMII) is a promising tool to help delineate the epileptogenic zone (EZ) in patients with medically intractable focal epilepsies undergoing presurgical evaluation. We report here the detailed methodology of MMII and an overview of the utility of MMII at the Cleveland Clinic Epilepsy Center from 2014 to 2018, exemplified by illustrative cases. The image integration was performed using the Curry platform (Compumedics Neuroscan™, Charlotte, NC, USA), including all available diagnostic modalities such as Magnetic resonance imaging (MRI), Positron Emission Tomography (PET), single-photon emission computed tomography (SPECT) and Magnetoencephalography (MEG), with additional capability of trajectory planning for intracranial EEG (ICEEG), particularly stereo-EEG (SEEG), as well as surgical resection planning. In the 5-year time span, 467 patients underwent MMII; of them, 98 patients (21%) had a history of prior neurosurgery and recurring seizures. Of the 467 patients, 425 patients underwent ICEEG implantation with further CT co-registration to identify the electrode locations. A total of 351 patients eventually underwent surgery after MMII, including 197 patients (56%) with non-lesional MRI and 223 patients (64%) with extra-temporal lobe epilepsy. Among 269 patients with 1-year post-operative follow up, 134 patients (50%) had remained completely seizure-free. The most common histopathological finding is focal cortical dysplasia. Our study illustrates the usefulness of MMII to enhance SEEG electrode trajectory planning, assist non-invasive/invasive data interpretation, plan resection strategy, and re-evaluate surgical failures. Information presented by MMII is essential to the understanding of the anatomo-functional-electro-clinical correlations in individual cases, which leads to the ultimate success of presurgical evaluation of patients with medically intractable focal epilepsies.
ABSTRACT
BACKGROUND AND OBJECTIVE: The purpose of the present study was to investigate low-shot deep learning models applied to conjunctival melanoma detection using a small dataset with ocular surface images. METHODS: A dataset was composed of anonymized images of four classes; conjunctival melanoma (136), nevus or melanosis (93), pterygium (75), and normal conjunctiva (94). Before training involving conventional deep learning models, two generative adversarial networks (GANs) were constructed to augment the training dataset for low-shot learning. The collected data were randomly divided into training (70%), validation (10%), and test (20%) datasets. Moreover, 3D melanoma phantoms were designed to build an external validation set using a smartphone. The GoogleNet, InceptionV3, NASNet, ResNet50, and MobileNetV2 architectures were trained through transfer learning and validated using the test and external validation datasets. RESULTS: The deep learning model demonstrated a significant improvement in the classification accuracy of conjunctival lesions using synthetic images generated by the GAN models. MobileNetV2 with GAN-based augmentation displayed the highest accuracy of 87.5% in the four-class classification and 97.2% in the binary classification for the detection of conjunctival melanoma. It showed an accuracy of 94.0% using 3D melanoma phantom images captured using a smartphone camera. CONCLUSIONS: The present study described a low-shot deep learning model that can detect conjunctival melanomas using ocular surface images. To the best of our knowledge, this study is the first to develop a deep learning model to detect conjunctival melanoma using a digital imaging device such as smartphone camera.
Subject(s)
Deep Learning , Melanoma , Nevus, Pigmented , Skin Neoplasms , Humans , Melanoma/diagnostic imaging , Neural Networks, ComputerABSTRACT
BACKGROUND: Dementia and cognitive impairment were significantly associated with hearing loss. The impact of hearing loss on dementia and cognitive impairment is understudied, particularly for different effect on cognitive impairment according to types of hearing loss. OBJECTIVE: The present study was conducted to elucidate the association between clinically diagnosed dementia and hearing loss with consideration of the type of hearing loss among an elderly population, and to explore the effects of different types of hearing loss on preclinical cognitive impairment. METHODS: Data (nâ=â59,675) from the Korean National Health Insurance Service-Health Screening were used to calculate odds ratios (OR) for cognitive impairment according to type of hearing loss (conductive, sensorineural, mixed, and noise-induced hearing losses, and presbycusis). Cognitive impairment was assessed using the Korean Dementia Screening Questionnaire-Prescreening (KDSQ-P). RESULTS: Cognitive impairment was significantly associated with conductive (OR: 1.45, 95% confidence interval (CI): 1.20-1.77), sensorineural (OR: 1.23, CI: 1.12-1.36), and noise-induced hearing loss (OR: 1.32, CI: 1.12-1.56), and presbycusis (OR: 1.53, CI: 1.25-1.87). Among participants scoring positive on the KDSQ-P (score≥4), the KDSQ-P score was significantly elevated in the mixed and noise-induced hearing loss groups. CONCLUSION: This study revealed a significant correlation between different types of hearing loss and cognitive impairment. Noise-induced hearing loss is especially important because it occurs earlier than other types of hearing loss and has large effects on cognitive impairment.
Subject(s)
Cognition/physiology , Cognitive Dysfunction/etiology , Dementia/etiology , Hearing Loss/complications , Adult , Aged , Disease Progression , Female , Humans , Male , Mass Screening , Middle Aged , Republic of KoreaABSTRACT
BACKGROUND: Recent studies have shown that long working hours can have adverse consequences on health and possibly trigger biological processes that mediate the relationship between long working hours and cognitive decline. OBJECTIVE: To investigate whether long working hours and the overall duration such exposure is associated with a decline in cognitive function. METHODS: Data obtained during the Korean Longitudinal Study on Aging (nâ=â2,518) during the period 2006-2018 were used to explore the relationship between long working hours and cognitive decline. Korean version of the Mini-Mental State Examination (K-MMSE) scores were used to evaluate cognitive function. Cox proportional hazard regression models were used to calculate hazard ratios (HRs) and 95% confidence intervals (CIs), which were used to evaluate declines in K-MMSE scores over the 12-year study period. RESULTS: Overall HR (95% CI) for a decline in cognitive function in long working hours group was 1.13 (0.73-1.17). When categorized by sex, women with long working hours had an HR (95% CI) of 1.50 (1.05-2.22), K-MMSE scores decreased significantly after working long hours for 5 years (pâ<â0.01). CONCLUSION: The study furthers understanding of the effects of long working hours on cognitive decline among female workers. Further research is required to determine the effects of long working hours on cognitive functions.
Subject(s)
Aging/physiology , Cognition/physiology , Cognitive Dysfunction/diagnosis , Sex Factors , Aged , Aged, 80 and over , Cognitive Dysfunction/physiopathology , Female , Follow-Up Studies , Humans , Korea , Longitudinal Studies , Male , Mental Status and Dementia Tests , Neuropsychological Tests , Proportional Hazards ModelsABSTRACT
BACKGROUND: To quantitatively evaluate the brain MRI morphological abnormalities in patients with cyclin-dependent kinase-like 5 deficiency disorder (CDD) on a group level and longitudinally. METHODS: We performed surface-based MRI analysis on high-resolution T1-weighted images on three CDD patients scanned at age of three years, and compared with 12 age- and gender-matched healthy controls. We further examined the longitudinal morphological changes in one patient with a follow-up of 5 years. RESULTS: CDD patients presented significant reductions in total intracranial volume, total gray matter (GM) volume and subcortical GM volume compared to controls. For subcortical regions, significant GM volume reductions were seen in the brain stem, bilateral thalamus, bilateral hippocampus, bilateral cerebellum and left amygdala. Although GM volume of cortical mantle did not show statistical differences overall, significant reduction was detected in bilateral parietal, left occipital and right temporal lobes. Cortical thickness exhibited significant decreases in bilateral occipital, parietal and temporal lobes, while surface area did not show any significant differences. Longitudinal follow-up in one patient revealed a monotonic downward trend of relative volume in the majority of brain regions. The relative surface area appeared to gain age-related growth, whereas the relative cortical thickness exhibited a striking progressive decline over time. CONCLUSIONS: Quantitative morphology analysis in children with CDD showed global volume loss in the cortex and more notably in the subcortical gray matter, with a progressive trend along with the disease course. Cortical thickness is a more sensitive measure to disclose cortical atrophy and disease progression than surface area.
Subject(s)
Brain/abnormalities , Brain/diagnostic imaging , Epileptic Syndromes/diagnostic imaging , Epileptic Syndromes/pathology , Spasms, Infantile/diagnostic imaging , Spasms, Infantile/pathology , Atrophy/diagnostic imaging , Atrophy/genetics , Atrophy/pathology , Child, Preschool , Disease Progression , Female , Humans , Infant , Magnetic Resonance Imaging/methods , MaleABSTRACT
Deep learning (DL) has been successfully applied to the diagnosis of ophthalmic diseases. However, rare diseases are commonly neglected due to insufficient data. Here, we demonstrate that few-shot learning (FSL) using a generative adversarial network (GAN) can improve the applicability of DL in the optical coherence tomography (OCT) diagnosis of rare diseases. Four major classes with a large number of datasets and five rare disease classes with a few-shot dataset are included in this study. Before training the classifier, we constructed GAN models to generate pathological OCT images of each rare disease from normal OCT images. The Inception-v3 architecture was trained using an augmented training dataset, and the final model was validated using an independent test dataset. The synthetic images helped in the extraction of the characteristic features of each rare disease. The proposed DL model demonstrated a significant improvement in the accuracy of the OCT diagnosis of rare retinal diseases and outperformed the traditional DL models, Siamese network, and prototypical network. By increasing the accuracy of diagnosing rare retinal diseases through FSL, clinicians can avoid neglecting rare diseases with DL assistance, thereby reducing diagnosis delay and patient burden.