Automatic Diagnosis of Bipolar Disorder Using Optical Coherence Tomography Data and Artificial Intelligence

Neuroretinal Alterations in Schizophrenia and Bipolar Disorder: An Updated Meta-analysis

Schizophrenia (SZ) and bipolar disorder (BD) are characterized by major symptomatic, cognitive, and neuroanatomical changes. Recent studies have used optical coherence tomography (OCT) to investigate retinal changes in SZ and BD, but their unique and shared changes require further evaluation. Articles were identified using PubMed and Google Scholar. 39 studies met the inclusion criteria. Diagnostic groups were proband (SZ/BD combined), SZ, BD, and healthy control (HC) eyes. Meta-analyses utilized fixed and random effects models when appropriate, and publication bias was corrected using trim-and-fill analysis ("meta" package in R). Results are reported as standardized mean differences with 95% CIs. Data from 3145 patient eyes (1956 SZ, 1189 BD) and 3135 HC eyes were included. Studies identified thinning of the peripapillary retinal nerve fiber layer (pRNFL, overall and in 2 subregions), m-Retina (overall and all subregions), mGCL-IPL, mIPL, and mRPE in SZ patients. BD showed thinning of the pRNFL (overall and in each subregion), pGCC, and macular Retina (in 5 subregions), but no changes in thickness or volume for the total retina. Neither SZ nor BD patients demonstrated significant changes in the fovea, mRNFL, mGCL, mGCC, mINL, mOPL, mONL, or choroid thicknesses. Moderating effects of age, illness duration, and smoking on retinal structures were identified. This meta-analysis builds upon previous literature in this field by incorporating recent OCT studies and examining both peripapillary and macular retinal regions with respect to psychotic disorders. Overall, this meta-analysis demonstrated both peripapillary and macular structural retinal abnormalities in people with SZ or BD compared with HCs.

Evaluation of Risk of Bias in Neuroimaging-Based Artificial Intelligence Models for Psychiatric Diagnosis: A Systematic Review

IMPORTANCE

Neuroimaging-based artificial intelligence (AI) diagnostic models have proliferated in psychiatry. However, their clinical applicability and reporting quality (ie, feasibility) for clinical practice have not been systematically evaluated.

OBJECTIVE

To systematically assess the risk of bias (ROB) and reporting quality of neuroimaging-based AI models for psychiatric diagnosis.

EVIDENCE REVIEW

PubMed was searched for peer-reviewed, full-length articles published between January 1, 1990, and March 16, 2022. Studies aimed at developing or validating neuroimaging-based AI models for clinical diagnosis of psychiatric disorders were included. Reference lists were further searched for suitable original studies. Data extraction followed the CHARMS (Checklist for Critical Appraisal and Data Extraction for Systematic Reviews of Prediction Modeling Studies) and PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-analyses) guidelines. A closed-loop cross-sequential design was used for quality control. The PROBAST (Prediction Model Risk of Bias Assessment Tool) and modified CLEAR (Checklist for Evaluation of Image-Based Artificial Intelligence Reports) benchmarks were used to systematically evaluate ROB and reporting quality.

FINDINGS

A total of 517 studies presenting 555 AI models were included and evaluated. Of these models, 461 (83.1%; 95% CI, 80.0%-86.2%) were rated as having a high overall ROB based on the PROBAST. The ROB was particular high in the analysis domain, including inadequate sample size (398 of 555 models [71.7%; 95% CI, 68.0%-75.6%]), poor model performance examination (with 100% of models lacking calibration examination), and lack of handling data complexity (550 of 555 models [99.1%; 95% CI, 98.3%-99.9%]). None of the AI models was perceived to be applicable to clinical practices. Overall reporting completeness (ie, number of reported items/number of total items) for the AI models was 61.2% (95% CI, 60.6%-61.8%), and the completeness was poorest for the technical assessment domain with 39.9% (95% CI, 38.8%-41.1%).

CONCLUSIONS AND RELEVANCE

This systematic review found that the clinical applicability and feasibility of neuroimaging-based AI models for psychiatric diagnosis were challenged by a high ROB and poor reporting quality. Particularly in the analysis domain, ROB in AI diagnostic models should be addressed before clinical application.

Objective Diagnosis of Fibromyalgia Using Neuroretinal Evaluation and Artificial Intelligence

Background/Objective

This study aims to identify objective biomarkers of fibromyalgia (FM) by applying artificial intelligence algorithms to structural data on the neuroretina obtained using swept-source optical coherence tomography (SS-OCT).

Method

The study cohort comprised 29 FM patients and 32 control subjects. The thicknesses of complete retina, 3 retinal layers [ganglion cell layer (GCL+), GCL++ (between the inner limiting membrane and the inner nuclear layer boundaries) and retinal nerve fiber layer (RNFL)] and choroid in 9 areas around the macula were obtained using SS-OCT. Discriminant capacity was evaluated using the area under the curve (AUC) and the Relief algorithm. A diagnostic aid system with an automatic classifier was implemented.

Results

No significant difference (p ≥ .660) was found anywhere in the choroid. In the RNFL, a significant difference was found in the inner inferior region (p = .010). In the GCL+, GCL++ layers and complete retina, a significant difference was found in the 4 regions defining the inner ring: temporal, superior, nasal and inferior. Applying an ensemble RUSBoosted tree classifier to the features with greatest discriminant capacity achieved accuracy = .82 and AUC = .82.

Conclusions

This study identifies a potential novel objective and non-invasive biomarker of FM based on retina analysis using SS-OCT.

Combined influence of medication and symptom severity on visual processing in bipolar disorder

Previous studies have reported visual impairments in patients with bipolar disorder (BPD), but unclear were whether clinical variables would be associated with those disturbances. Here, we investigate the relationship between visual functioning, in terms of color discrimination, and the impact of BPD duration, mood state, and the patients' medication. Forty-five participants (25-45 years old) were recruited for this study. Color discrimination was performed using the Cambridge Colour Test. Serial multiple mediations were run to investigate the assumption of association between color discrimination and the clinical variables. Our findings showed that, compared with healthy controls, BPD patients' performance was worse for the Protan, Deutan, and Tritan vectors, revealing deterioration of color discrimination. In addition, the mediation analyses revealed a strong direct (p < .001) and moderate-to-high indirect effects (p < .01) of medication and symptom severity on color discrimination. Overall, both longer the duration of the disease and greater the symptom severity of BPD patients resulted in worse performance. It highlights the importance of examining the wider clinical context of an affective disorder to understand how it affects visual processing in this population.

Development of a Novel Retina-Based Diagnostic Score for Early Detection of Major Depressive Disorder: An Interdisciplinary View

BACKGROUND

Clinically effective markers for the diagnosis of major depressive disorder (MDD) are lacking. Alterations in retinal features are closely related to the pathophysiological progression of MDD. However, the reliable retina-related diagnostic model for MDD remains to be developed. Thus, our study aimed to quantitatively evaluate retinal vascular and structural changes in MDD patients and to develop a reliable diagnostic model of MDD based on retinal parameters.

METHODS

Seventy-eight patients with MDD and 47 healthy controls (HCs) underwent retinal vessel density and structure examination using optical coherence tomography angiography and visual field examination using perimetry. Independent-sample t test was used to assess the differences in retinal parameters between the groups. Meanwhile, we constructed the corresponding retina-based diagnostic model by LASSO logistic regression. Finally, the diagnostic ability of the model was evaluated by area under the curve (AUC) of receiver operating characteristic curves and calibration plot of nomogram.

RESULTS

MDD patients showed lower retinal vessel density (including radial peripapillary capillary vessel density, superficial and deep capillary plexus vessel density), thinner subfoveal choroidal thickness, and poorer visual fields compared to HCs (all p < 0.05). Furthermore, a retina-based diagnostic model was constructed and shows a strong diagnostic capability for MDD (AUC = 0.9015, p < 0.001).

CONCLUSION

Patients with MDD showed distinct retinal features compared to HCs. The retina-based diagnostic model is expected to be a necessary complement to the diagnosis of MDD.

Early Diagnosis of Multiple Sclerosis Using Swept-Source Optical Coherence Tomography and Convolutional Neural Networks Trained with Data Augmentation

BACKGROUND

The aim of this paper is to implement a system to facilitate the diagnosis of multiple sclerosis (MS) in its initial stages. It does so using a convolutional neural network (CNN) to classify images captured with swept-source optical coherence tomography (SS-OCT).

METHODS

SS-OCT images from 48 control subjects and 48 recently diagnosed MS patients have been used. These images show the thicknesses (45 × 60 points) of the following structures: complete retina, retinal nerve fiber layer, two ganglion cell layers (GCL+, GCL++) and choroid. The Cohen distance is used to identify the structures and the regions within them with greatest discriminant capacity. The original database of OCT images is augmented by a deep convolutional generative adversarial network to expand the CNN's training set.

RESULTS

The retinal structures with greatest discriminant capacity are the GCL++ (44.99% of image points), complete retina (26.71%) and GCL+ (22.93%). Thresholding these images and using them as inputs to a CNN comprising two convolution modules and one classification module obtains sensitivity = specificity = 1.0.

CONCLUSIONS

Feature pre-selection and the use of a convolutional neural network may be a promising, nonharmful, low-cost, easy-to-perform and effective means of assisting the early diagnosis of MS based on SS-OCT thickness data.