Deep Learning-Based Automated Detection of Retinal Breaks and Detachments on Fundus Photography

Purpose

The purpose of this study was to develop a deep learning algorithm, to detect retinal breaks and retinal detachments on ultra-widefield fundus (UWF) optos images using artificial intelligence (AI).

Methods

Optomap UWF images of the database were annotated to four groups by two retina specialists: (1) retinal breaks without detachment, (2) retinal breaks with retinal detachment, (3) retinal detachment without visible retinal breaks, and (4) a combination of groups 1 to 3. The fundus image data set was split into a training set and an independent test set following an 80% to 20% ratio. Image preprocessing methods were applied. An EfficientNet classification model was trained with the training set and evaluated with the test set.

Results

A total of 2489 UWF images were included into the dataset, resulting in a training set size of 2008 UWF images and a test set size of 481 images. The classification models achieved an area under the receiver operating characteristic curve (AUC) on the testing set of 0.975 regarding lesion detection, an AUC of 0.972 for retinal detachment and an AUC of 0.913 for retinal breaks.

Conclusions

A deep learning system to detect retinal breaks and retinal detachment using UWF images is feasible and has a good specificity. This is relevant for clinical routine as there can be a high rate of missed breaks in clinics. Future clinical studies will be necessary to evaluate the cost-effectiveness of applying such an algorithm as an automated auxiliary tool in a large practices or tertiary referral centers.

Translational Relevance

This study demonstrates the relevance of applying AI in diagnosing peripheral retinal breaks in clinical routine in UWF fundus images.

Extent of cord pathology in the lumbosacral enlargement in non-traumatic versus traumatic spinal cord injury

This study compares remote neurodegenerative changes caudal to a cervical injury in degenerative cervical myelopathy (DCM) (i.e., non-traumatic) and incomplete traumatic spinal cord injury (tSCI) patients, using MRI-based tissue area measurements and diffusion tensor imaging (DTI). Eighteen mild to moderate DCM patients with sensory impairments (mJOA score: 16.2±1.9), 14 incomplete tetraplegic tSCI patients (AIS C&D), and 20 healthy controls were recruited. All participants received DTI and T2*-weighted scans in the lumbosacral enlargement (caudal to injury) and at C2/C3 (rostral to injury). MRI readouts included DTI metrics in the white matter (WM) columns and cross-sectional WM and gray matter area. One-way ANOVA with Tukey's post-hoc comparison (p<0.05) was used to assess group differences. In the lumbosacral enlargement, compared to DCM, tSCI patients exhibited decreased fractional anisotropy in the lateral (tSCI vs. DCM, -11.9%, p=0.007) and ventral WM column (-8.0%, p=0.021), and showed trend toward lower values in the dorsal column (-8.9%, p=0.068). At C2/C3, compared to controls, fractional anisotropy was lower in both groups in the dorsal (DCM vs. controls, -7.9%, p=0.024; tSCI vs. controls, -10.0%, p=0.007) and in the lateral column (DCM: -6.2%, p=0.039; tSCI: -13.3%, p<0.001), while tSCI patients had lower fractional anisotropy than DCM patients in the lateral column (-7.6%, p=0.029). WM areas were not different between patient groups but were lower compared to controls in the lumbosacral enlargement (DCM: -16.9%, p<0.001; tSCI, -10.5%, p=0.043) and at C2/C3 (DCM: -16.0%, p<0.001; tSCI: -18.1%, p<0.001). In conclusion, mild to moderate DCM and incomplete tSCI lead to similar degree of degeneration of the dorsal and lateral columns at C2/C3, but tSCI results in more widespread white matter damage in the lumbosacral enlargement. These remote changes are likely to contribute to the patients' impairment and recovery. DTI is a sensitive tool to assess remote pathological changes in DCM and tSCI patients.

Comparison of axial and sagittal spinal cord motion measurements in degenerative cervical myelopathy

BACKGROUND AND PURPOSE

The timing of decision-making for a surgical intervention in patients with mild degenerative cervical myelopathy (DCM) is challenging. Spinal cord motion phase contrast MRI (PC-MRI) measurements can reveal the extent of dynamic mechanical strain on the spinal cord to potentially identify high-risk patients. This study aims to determine the comparability of axial and sagittal PC-MRI measurements of spinal cord motion with the prospect of improving the clinical workup.

METHODS

Sixty-four DCM patients underwent a PC-MRI scan assessing spinal cord motion. The agreement of axial and sagittal measurements was determined by means of intraclass correlation coefficients (ICCs) and Bland-Altman analyses.

RESULTS

The comparability of axial and sagittal PC-MRI measurements was good to excellent at all cervical levels (ICCs motion amplitude: .810-.940; p &lt; .001). Significant differences between axial and sagittal amplitude values could be found at segments C3 and C4, while its magnitude was low (C3: 0.07 ± 0.19 cm/second; C4: -0.12 ± 0.30 cm/second). Bland-Altman analysis showed a good agreement between axial and sagittal PC-MRI scans (coefficients of repeatability: minimum -0.23 cm/second at C2; maximum -0.58 cm/second at C4). Subgroup analysis regarding anatomic conditions (stenotic vs. nonstenotic segments) and different velocity encoding (2 vs. 3 cm/second) showed comparable results.

CONCLUSIONS

This study demonstrates good comparability between axial and sagittal spinal cord motion measurements in DCM patients. To this end, axial and sagittal PC-MRI are both accurate and sensitive in detecting pathologic cord motion. Therefore, such measures could identify high-risk patients and improve clinical decision-making (ie, timing of decompression).

Longitudinal changes of spinal cord grey and white matter following spinal cord injury

OBJECTIVES

Traumatic and non-traumatic spinal cord injury produce neurodegeneration across the entire neuraxis. However, the spatiotemporal dynamics of spinal cord grey and white matter neurodegeneration above and below the injury is understudied.

METHODS

We acquired longitudinal data from 13 traumatic and 3 non-traumatic spinal cord injury patients (8-8 cervical and thoracic cord injuries) within 1.5 years after injury and 10 healthy controls over the same period. The protocol encompassed structural and diffusion-weighted MRI rostral (C2/C3) and caudal (lumbar enlargement) to the injury level to track tissue-specific neurodegeneration. Regression models assessed group differences in the temporal evolution of tissue-specific changes and associations with clinical outcomes.

RESULTS

At 2 months post-injury, white matter area was decreased by 8.5% and grey matter by 15.9% in the lumbar enlargement, while at C2/C3 only white matter was decreased (-9.7%). Patients had decreased cervical fractional anisotropy (FA: -11.3%) and increased radial diffusivity (+20.5%) in the dorsal column, while FA was lower in the lateral (-10.3%) and ventral columns (-9.7%) of the lumbar enlargement. White matter decreased by 0.34% and 0.35% per month at C2/C3 and lumbar enlargement, respectively, and grey matter decreased at C2/C3 by 0.70% per month.

CONCLUSIONS

This study describes the spatiotemporal dynamics of tissue-specific spinal cord neurodegeneration above and below a spinal cord injury. While above the injury, grey matter atrophy lagged initially behind white matter neurodegeneration, in the lumbar enlargement these processes progressed in parallel. Tracking trajectories of tissue-specific neurodegeneration provides valuable assessment tools for monitoring recovery and treatment effects.

Tracking White and Gray Matter Degeneration along the Spinal Cord Axis in Degenerative Cervical Myelopathy

This study aims to determine tissue-specific neurodegeneration across the spinal cord in patients with mild-moderate degenerative cervical myelopathy (DCM). Twenty-four mild-moderate DCM and 24 healthy subjects were recruited. In patients, a T2-weighted scan was acquired at the compression site, whereas in all participants a T2*-weighted and diffusion-weighted scan was acquired at the cervical level (C2-C3) and in the lumbar enlargement (i.e., rostral and caudal to the site of compression). We quantified intramedullary signal changes, maximal canal and cord compression, white (WM) and gray matter (GM) atrophy, and microstructural indices from diffusion-weighted scans. All patients underwent clinical (modified Japanese Orthopaedic Association; mJOA) and electrophysiological assessments. Regression analysis assessed associations between magnetic resonance imaging (MRI) readouts and electrophysiological and clinical outcomes. Twenty patients were classified with mild and 4 with moderate DCM using the mJOA scale. The most frequent site of compression was at the C5-C6 level, with maximum cord compression of 38.73% ± 11.57%. Ten patients showed imaging evidence of cervical myelopathy. In the cervical cord, WM and GM atrophy and WM microstructural changes were evident, whereas in the lumbar cord only WM showed atrophy and microstructural changes. Remote cervical cord WM microstructural changes were pronounced in patients with radiological myelopathy and associated with impaired electrophysiology. Lumbar cord WM atrophy was associated with lower limb sensory impairments. In conclusion, tissue-specific neurodegeneration revealed by quantitative MRI is already apparent across the spinal cord in mild-moderate DCM before the onset of severe clinical impairments. WM microstructural changes are particularly sensitive to remote pathologically and clinically eloquent changes in DCM.

The Restless Spinal Cord in Degenerative Cervical Myelopathy

BACKGROUND AND PURPOSE

The spinal cord is subject to a periodic, cardiac-related movement, which is increased at the level of a cervical stenosis. Increased oscillations may exert mechanical stress on spinal cord tissue causing intramedullary damage. Motion analysis thus holds promise as a biomarker related to disease progression in degenerative cervical myelopathy. Our aim was characterization of the cervical spinal cord motion in patients with degenerative cervical myelopathy.

MATERIALS AND METHODS

Phase-contrast MR imaging data were analyzed in 55 patients (37 men; mean age, 56.2 [SD,12.0] years; 36 multisegmental stenoses) and 18 controls (9 men, P = .368; mean age, 62.2 [SD, 6.5] years; P = .024). Parameters of interest included the displacement and motion pattern. Motion data were pooled on the segmental level for comparison between groups.

RESULTS

In patients, mean craniocaudal oscillations were increased manifold at any level of a cervical stenosis (eg, C5 displacement: controls [n = 18], 0.54 [SD, 0.16] mm; patients [n = 29], monosegmental stenosis [n = 10], 1.86 [SD, 0.92] mm; P < .001) and even in segments remote from the level of the stenosis (eg, C2 displacement: controls [n = 18], 0.36 [SD, 0.09] mm; patients [n = 52]; stenosis: C3, n = 21; C4, n = 11; C5, n = 18; C6, n = 2; 0.85 [SD, 0.46] mm; P < .001). Motion at C2 differed with the distance to the next stenotic segment and the number of stenotic segments. The motion pattern in most patients showed continuous spinal cord motion throughout the cardiac cycle.

CONCLUSIONS

Patients with degenerative cervical myelopathy show altered spinal cord motion with increased and ongoing oscillations at and also beyond the focal level of stenosis. Phase-contrast MR imaging has promise as a biomarker to reveal mechanical stress to the cord and may be applicable to predict disease progression and the impact of surgical interventions.

Predictive Value of Midsagittal Tissue Bridges on Functional Recovery After Spinal Cord Injury

Background

The majority of patients with spinal cord injury (SCI) have anatomically incomplete lesions and present with preserved tissue bridges, yet their outcomes vary.

Objective

To assess the predictive value of the anatomical location (ventral/dorsal) and width of preserved midsagittal tissue bridges for American Spinal Injury Association (ASIA) Impairment Scale (AIS) grade conversion and SCI patient stratification into recovery-specific subgroups.

Methods

This retrospective longitudinal study includes 70 patients (56 men, age: 52.36 ± 18.58 years) with subacute (ie, 1 month) SCI (45 tetraplegics, 25 paraplegics), 1-month neuroimaging data, and 1-month and 12-month clinical data. One-month midsagittal T2-weighted scans were used to determine the location and width of tissue bridges. Their associations with functional outcomes were assessed using partial correlation and unbiased recursive partitioning conditional inference tree (URP-CTREE).

Results

Fifty-seven (81.4%) of 70 patients had tissue bridges (2.53 ± 2.04 mm) at 1-month post-SCI. Larger ventral (P = .001, r = 0.511) and dorsal (P < .001, r = 0.546) tissue bridges were associated with higher AIS conversion rates 12 months post-SCI (n = 39). URP-CTREE analysis identified 1-month ventral tissue bridges as predictors of 12-month total motor scores (0.4 mm cutoff, P = .008), recovery of upper extremity motor scores at 12 months (1.82 mm cutoff, P = .002), 12-month pin-prick scores (1.4 mm cutoff, P = .018), and dorsal tissue bridges at 1 month as predictors of 12-month Spinal Cord Independence Measure scores (0.5 mm cutoff, P = .003).

Conclusions

Midsagittal tissue bridges add predictive value to baseline clinical measures for post-SCI recovery. Based on tissue bridges’ width, patients can be classified into subgroups of clinical recovery profiles. Midsagittal tissue bridges provide means to optimize patient stratification in clinical trials.

Tissue bridges predict neuropathic pain emergence after spinal cord injury

OBJECTIVE

To assess associations between preserved spinal cord tissue quantified by the width of ventral and dorsal tissue bridges and neuropathic pain development after spinal cord injury.

METHODS

This retrospective longitudinal study includes 44 patients (35 men; mean (SD) age, 50.05 (18.88) years) with subacute (ie, 1 month) spinal cord injury (25 patients with neuropathic pain, 19 pain-free patients) and neuroimaging data who had a follow-up clinical assessment at 12 months. Widths of tissue bridges were calculated from midsagittal T2-weighted images and compared across groups. Regression analyses were used to identify relationships between these neuroimaging measures and previously assessed pain intensity and pin-prick score.

RESULTS

Pin-prick score of the 25 patients with neuropathic pain increased from 1 to 12 months (Δmean=10.08, 95% CI 2.66 to 17.50, p=0.010), while it stayed similar in pain-free patients (Δmean=2.74, 95% CI -7.36 to 12.84, p=0.576). They also had larger ventral tissue bridges (Δmedian=0.80, 95% CI 0.20 to 1.71, p=0.008) at 1 month when compared with pain-free patients. Conditional inference tree analysis revealed that ventral tissue bridges' width (≤2.1 or >2.1 mm) at 1 month is the strongest predictor for 12 months neuropathic pain intensity (1.90±2.26 and 3.83±1.19, p=0.042) and 12 months pin-prick score (63.84±28.26 and 92.67±19.43, p=0.025).

INTERPRETATION

Larger width of ventral tissue bridges-a proxy for spinothalamic tract function-at 1 month post-spinal cord injury is associated with the emergence and maintenance of neuropathic pain and increased pin-prick sensation. Spared ventral tissue bridges could serve as neuroimaging biomarkers of neuropathic pain and might be used for prediction and monitoring of pain outcomes and stratification of patients in interventional trials.

Cervical Cord Neurodegeneration in Traumatic and Non-Traumatic Spinal Cord Injury

This study aimed to compare macrostructural and microstructural neurodegenerative changes remote from a cervical spinal cord injury in traumatic spinal cord injury (tSCI) and degenerative cervical myelopathy (DCM) patients using quantitative magnetic resonance imaging (MRI). Twenty-nine tSCI patients, 20 mild/moderate DCM patients, and 22 healthy controls underwent a high-resolution MRI protocol at the cervical cord (C2/C3). High-resolution T2*-weighted and diffusion-weighted scans provided data to calculate tissue-specific cross-sectional areas of the spinal cord and tract-specific diffusion indices of cord white matter, respectively. Regression analysis determined associations between neurodegeneration and clinical impairment. tSCI patients showed more impairment in upper limb strength and manual dexterity when compared with DCM patients. While macrostructural MRI measures revealed a similar extent of remote cord atrophy at cervical level, microstructural measures (diffusion indices) were able to distinguish more pronounced tract-specific neurodegeneration in tSCI patients when compared with DCM patients. Tract-specific neurodegeneration was associated with upper limb impairment. Despite clinical differences between severely impaired tSCI compared with mildly affected DCM patient, extensive cord atrophy is present remotely from the focal spinal cord injury. Diffusion indices revealed greater tract-specific alterations in tSCI patients. Therefore, diffusion indices are more sensitive than macrostructural MRI measures as these are able to distinguish between traumatic and non-traumatic spinal cord injury. Neuroimaging biomarkers of cervical cord integrity hold potential as predictors of recovery and might be suitable biomarkers for interventional trials both in traumatic and non-traumatic SCI.

Width and neurophysiologic properties of tissue bridges predict recovery after cervical injury

OBJECTIVE

To assess whether preserved dorsal and ventral midsagittal tissue bridges after traumatic cervical spinal cord injury (SCI) encode tract-specific electrophysiologic properties and are predictive of appropriate recovery.

METHODS

In this longitudinal study, we retrospectively assessed MRI scans at 1 month after SCI that provided data on width and location (dorsal vs ventral) of midsagittal tissue bridges in 28 tetraplegic patients. Regression analysis assessed associations between midsagittal tissue bridges and motor- and sensory-specific electrophysiologic recordings and appropriate outcome measures at 12 months after SCI.

RESULTS

Greater width of dorsal midsagittal tissue bridges at 1 month after SCI identified patients who were classified as being sensory incomplete at 12 months after SCI (p = 0.025), had shorter sensory evoked potential (SEP) latencies (r = -0.57, p = 0.016), and had greater SEP amplitudes (r = 0.61, p = 0.001). Greater width of dorsal tissue bridges predicted better light-touch score at 12 months (r = 0.40, p = 0.045) independently of baseline clinical score and ventral tissue bridges. Greater width of ventral midsagittal tissue bridges at 1 month identified patients who were classified as being motor incomplete at 12 months (p = 0.002), revealed shorter motor evoked potential (MEP) latencies (r = -0.54, p = 0.044), and had greater ratios of MEP amplitude to compound muscle action potential amplitude (r = 0.56, p = 0.005). Greater width of ventral tissue bridges predicted better lower extremity motor scores at 12 months (r = 0.41, p = 0.035) independently of baseline clinical score and dorsal tissue bridges.

CONCLUSION

Midsagittal tissue bridges, detectable early after SCI, underwrite tract-specific electrophysiologic communication and are predictors of appropriate sensorimotor recovery. Neuroimaging biomarkers of midsagittal tissue bridges may be integrated into the diagnostic workup, prediction of recovery, and patients' stratification in clinical trials.

[Evaluation of an outpatient multidisciplinary Neuro-HIV clinic by the patients and referring doctors]

The neurocognitive complaints among HIV infected patients remain frequent, and to establish their etiology can be challenging. We created in 2011 an outpatient Neuro-HIV clinical platform that takes advantage of a multidisciplinary approach with 5 specialists (neuropsychologist, neurologist, psychiatrist, infectiologist and neuroradiologist). In order to estimate its utility, we conducted two questionnaire-based interviews by phone calls with the patients and their referring physicians. Three quarters of both the patients and the physicians interviewed consider the platform useful or essential. Even though there is often no immediate treatment for cognitive disorders, the patients receive from this multidisciplinary approach a better understanding of their disease, which may help them to better cope with their anxieties in daily life.