Predicting visual acuity with machine learning in treated ocular trauma patients.

AIM: To predict best-corrected visual acuity (BCVA) by machine learning in patients with ocular trauma who were treated for at least 6mo. METHODS: The internal dataset consisted of 850 patients with 1589 eyes and an average age of 44.29y. The initial visual acuity was 0.99 logMAR. The test dataset consisted of 60 patients with 100 eyes collected while the model was optimized. Four different machine-learning algorithms (Extreme Gradient Boosting, support vector regression, Bayesian ridge, and random forest regressor) were used to predict BCVA, and four algorithms (Extreme Gradient Boosting, support vector machine, logistic regression, and random forest classifier) were used to classify BCVA in patients with ocular trauma after treatment for 6mo or longer. Clinical features were obtained from outpatient records, and ocular parameters were extracted from optical coherence tomography images and fundus photographs. These features were put into different machine-learning models, and the obtained predicted values were compared with the actual BCVA values. The best-performing model and the best variable selected were further evaluated in the test dataset. RESULTS: There was a significant correlation between the predicted and actual values [all Pearson correlation coefficient (PCC)>0.6]. Considering only the data from the traumatic group (group A) into account, the lowest mean absolute error (MAE) and root mean square error (RMSE) were 0.30 and 0.40 logMAR, respectively. In the traumatic and healthy groups (group B), the lowest MAE and RMSE were 0.20 and 0.33 logMAR, respectively. The sensitivity was always higher than the specificity in group A, in contrast to the results in group B. The classification accuracy and precision were above 0.80 in both groups. The MAE, RMSE, and PCC of the test dataset were 0.20, 0.29, and 0.96, respectively. The sensitivity, precision, specificity, and accuracy of the test dataset were 0.83, 0.92, 0.95, and 0.90, respectively. CONCLUSION: Predicting BCVA using machine-learning models in patients with treated ocular trauma is accurate and helpful in the identification of visual dysfunction.

[Forensic Application of Microperimetry and Visual Evoked Potential in Macular Disease].

OBJECTIV: e To find the correlation between real best corrected visual acuity (BCVA) and testing results of microperimetry and visual evoked potential (VEP) and to explore a new method in recording BCVA in macular disease. METHODS: Sixty-two patients with macular disease (macular disease group, 62 eyes) and eighteen healthy volunteers (control group, 36 eyes) had BCVA, microperimetry and VEP recorded. RESULTS: (1) By microperimetry, the values of retinal mean sensitivity and fixation percentage in macular disease group were lower than that in control group. The bicurve ellipse area in macular disease group was higher than that in control group. By VEP, P100 amplitude under 0.5 cpd and 2 cpd in macular disease group were significantly higher than that in control group and the latency was prolonged (P < 0.05). (2) In macular disease group, BCVA had significant positive correlation with retinal mean sensitivity, bicurve ellipse area, macular central 2 degrees and 4 degrees fixation percentage, respectively (P < 0.05). There was a significant correlation between retinal mean sensitivity and P100 amplitude (P < 0.05). (3) Multiple linear regression equation was y = 0.053 x1+0.008 x3+3.897 (y was BCVA, while x1 was retinal mean sensitivity and x3 was P100 amplitude under 2 cpd). CONCLUSION: Combined use of microperimetry and VEP is useful in the assessment of BCVA in macular disease.

[Comparison of anterior chamber angle examination by UBM, SL-OCT and gonioscopy].

OBJECTIVE: To compare the agreement of anterior chamber angle examination by ultrasound biomicroscope (UBM), slit lamp optical coherence tomography (SL-OCT), and gonioscopy in angle recession and angle closure. METHODS: The anterior chamber angle was measured with UBM, SL-OCT and gonioscopy in turns for temporal, nasal, superior and inferior quadrant in the same dark room. The results were compared with the agreement of the three methods in angle recession and angle closure by χ2 test and Kappa test. RESULTS: There were no statistically significant differences of the three methods in testing angle closure and angle recession (P>0.05). The consistency of UBM and gonioscopy was better (Kappa value of 0.882) than that of SL-OCT and gonioscopy (Kappa value of 0.624). CONCLUSION: When testing angle recession, UBM is better than SL-OCT with gonioscopy as the standard. When testing angle closure, UBM, SL-OCT and gonioscopy have good agreement.

[Multifocal visual electrophysiology in visual function evaluation].

Multifocal visual electrophysiology, consisting of multifocal electroretinography (mfERG) and multifocal visual evoked potential (mfVEP), can objectively evaluate retina function and retina-cortical conduction pathway status by stimulating many local retinal regions and obtaining each local response simultaneously. Having many advantages such as short testing time and high sensitivity, it has been widely used in clinical ophthalmology, especially in the diagnosis of retinal disease and glaucoma. It is a new objective technique in clinical forensic medicine involving visual function evaluation of ocular trauma in particular. This article summarizes the way of stimulation, the position of electrodes, the way of analysis, the visual function evaluation of mfERG and mfVEP, and discussed the value of multifocal visual electrophysiology in forensic medicine.

[Application of contrast vision in identifying the malingering decreased vision].

OBJECTIVE: To study the application value of contrast vision in identifying the malingering decreased vision in the practice of clinical forensic medicine. METHODS: Thirty-one young and middle aged volunteers were selected and divided randomly into group 1 (16 persons with 32 eyes) and group 2 (15 persons with 30 eyes). The optotype contrast was 100%, 25%, 10% and 5%, respectively. The contrast vision of group 1 was tested. The contrast vision of group 2 was tested as follows: (1) the volunteers cooperated without inspector's interference; (2) the volunteers cooperated under inspector's interference; (3) the volunteers disguised decreased vision without inspector's interference; (4) the volunteers disguised decreased vision under inspector's interference. The data was then analyzed by statistics. RESULTS: There was a close correlation between contrast vision and contrast. As the contrast decreased, the vision acuity also decreased. The contrast vision curve of former two methods showed a good reproducibility while the contrast vision curve of latter two methods had a bad reproducibility. CONCLUSION: The repetition of contrast curve with or without inspector's interference can be used to discriminate malingering vision. The acquired contrast curves can be provided to the court as direct evidence and can help enhance the verification conclusion.

[The best corrected presenting distance visual acuity in forensic medicine].

At present the sight impairment evaluation in forensic medicine of China is based on the international classification of disease by WHO in 1973. The main measured indicator is "best corrected visual acuity". It is different from "presenting distance visual acuity" in some situations. In the new blindness and vision loss classification made by WHO in 2003, "presenting distance visual acuity" took the place of the "best corrected visual acuity". In the practice of forensic medicine, "presenting distance visual acuity" can not reflect the real visual acuity duo to the exaggeration or disguise of the wounded. We suggest to use "the best corrected presenting distance visual acuity" instead of "presenting distance visual acuity" in order to avoid the influences of the exaggeration or disguise of the wounded.

[The relationship between eyeball structure and visual acuity in high myopia].

OBJECTIVE: To explore the relationship between eyeball structure and visual acuity in high myopia. METHODS: Totally, 152 people (283 eyeballs) with different levels of myopia were tested for visual acuity, axial length, and fundus. All cases were classified according to diopter, axial length, and fundus. The relationships between diopter, axial length, fundus and visual acuity were studied. The mathematical models were established for visual acuity and eyeball structure markers. RESULTS: The visual acuity showed a moderate correlation with fundus class, comus, axial length and diopter ([r] > 0.4, P < 0.000 1). The visual acuity in people with the axial length longer than 30.00 mm, diopter above -20.00 D and fundus in 4th class were mostly below 0.5. The mathematical models were established by visual acuity and eyeball structure markers. CONCLUSION: The visual acuity should decline with axial length extension, diopter deepening and pathological deterioration of fundus. To detect the structure changes by combining different kinds of objective methods can help to assess and to judge the vision in high myopia.

[Image of the traumatic optic atrophy detected with optical coherence tomography and the vision function defection].

OBJECTIVE: To analyze the unilateral image of the traumatic optic nerve atrophy using optical coherence tomography (OCT) and to explore the relationship between the optic nerve atrophy and visual acuity or visual field. METHODS: Fifteen cases with traumatic optic nerve atrophy were studied. All cases were tested by routine opthalmological examination, visual evoked potentials (VEP) and optic disc examination with OCT. In cases with visual acuity more than 0.1 the visual field was tested. RESULTS: All cases had visual acuity and visual field defect at various levels and optic disc retinal nerve fiber layer (RNFL) atrophy. CONCLUSION: The traumatic optic atrophy begins in the temporal area of optic disc. The nasal area's atrophy occurs at the last and is the mildest. The more serious is the optic nerve atrophy, the worse is the vision function impairment. OCT is a valuable technique for evaluating the optic nerve atrophy and has good correlation with the visual function.

[Effects of temporal frequency on the sweep pattern visual evoked response acuity].

OBJECTIVE: To investigate the changes of sweep pattern visual evoked response (SPVER) acuity in different temporal frequency. METHODS: The persons were collected and divided into the low visual acuity group (71 eyes) and the high visual acuity group (92 eyes). The eyes were stimulated with different temporal frequency (8.5 Hz, 10 Hz, 15 Hz and 20 Hz, respectively) and the same series of spatial frequency. The accuracy of SPVER acuity in different temporal frequency was evaluated by compare the SPVER acuity with the subjective vision. RESULTS: In the low visual acuity group, there was a good correlation between the SPVER acuity and the subjective vision applying the spatial frequency of the group. In the high visual acuity group, the SPVER acuity was lower than the subjective vision. SPVER acuity showed a good correlation with the subjective vision at the temporal frequency of 10 Hz in two groups. CONCLUSION: The temporal frequency can influence SPVER accuracy, so it is necessary to select the suitable stimulation condition in forensic practice.

[Eyeball structure changes in high myopic patients and their significance for forensic assessment].

There are irreversible eyeball structural changes in high myopic patients. These changes include axial length, corneal radius, anterior chamber depth, fundus degeneration, macula thickness, etc. There is a close relationship between the damage degree of visual function and these changes. The incidence of complications, such as vitreous opacity, posterior vitreous detachment, cataract, glaucoma, posterior staphyloma and retina detachment, is also highly related to the myopia diopter. More and more researches have indicated that the myopia diopter and the level of visual function are affected by multiple factors. It is promising to detect all of these changes by different kinds of methods, and to assess visual function through these changes. By clarifying these changes, it is also useful to distinguish traumatic damage from disease to provide evidence for forensic assessment of eye injuries.

[Accurate determination of wound depth with ultrasound and its forensic significance].

Accurate determination of the wound depth has long been a difficult task in forensic medicine due to lack of technology. Nowadays, the depth of the wound can be accurately determined by using high frequency 2-D ultrasound and Color Doppler Flow Imaging (CDFI). Two typical cases with ultrasonic images were reported to show the importance of ultrasound technology in forensic medicine. The new technology provides scientific evidence of the injury assessment.

[Age estimation by dental radiological imaging].

Dental age plays an important role in age estimation. It has often been used together with skeletal age to improve the accuracy of age estimation abroad, but seldom performed in China. As a noninvasive technology, dental radiological imaging has been widely used in age estimation. By observing the age-related changes such as the pulp cavity and development of crown and root on radiographs. Gleiser and Hunt, as well as Demirjian have developed different methods to determine the age of human. Demirjian's method has been proved to be more accurate but with limitation when used in persons of eighteen and above. The accuracy and reliability of the measurements on pulp cavity could be improved as the development of computed tomography with its high resolution and intelligent software. As a convenient and accurate method, age estimation from dental computed tomographs would be more promising in the future for forensic scientists and anthropologists.

[Accuracy of age estimation from orthopantomograph using Demirjian's method].

OBJECTIVE: Developing teeth are used to assess maturity and estimate age in a number of disciplines. The aim of this study was to determine the accuracy of Demirjian method (Panoramic radiographs were examined and seven mandibular teeth staged according to Demirjian's dental maturity scale) in the forensic clinical medicine. METHODS: Tooth formation was assessed with orthopantomographs in healthy children in dental teaching hospital. There were total 828 children, with 279 boys and 549 girls, aged from 11 to 19 years. The difference between dental and real age was compared and measured, using t-test. RESULTS: The Demirjian method overestimated age in the aged 11-16 years group and had limitations in aged group over 17 years. CONCLUSION: The 95% confidence interval of the mean was least for mean of all developing teeth using Demirjian method (age 11-16 years).

[Forensic medical assessment in post-trauma retinal detachment].

OBJECTIVE: To analyze risk factors associated with post-traumatic retinal detachment and to identify the cause-effect relationship between retinal detachment and blunt ocular trauma in forensic medical assessment. METHODS: 112 cases of forensic medical examination on post -traumatic retinal detachment were retrospectively analyzed. RESULTS: A variety of retinal abnormalities were found in these cases, including tearing of the ora serrata retinae (4.28%), macular perforation (12.50%), small (<90 degrees) nonmacular retinal perforation (56.25%), larger (>90 degrees) retinal perforation (5.00%), and tractive retinal detachment without perforation (11.6%). Proliferative vitreoretinopathy (PVR) was found in 45 eyes examined. In most cases, retinal detachment occurred between 1 week to 2 months after blunt ocular trauma (61.60%), with 83.93% accompanied with severe myopia (>-3.00D) and 52.67% accompanied with vitreous floaters. Of all cases, 41.07% were directly and 52.68% were indirectly resulted from blunt ocular trauma, and the rest (6.25%) showed no association with blunt ocular trauma. CONCLUSION Many risk factors may result in retinal detachment including blunt ocular trauma and other causes. Accurate assessment of the relationship between blunt ocular trauma and retinal detachment is an important part of forensic examination.

[Comparison of objective assessment using the sweep pattern visual evoked response acuity (SPVERA) and illiterate E visual acuity].

OBJECTIVE: To compare the vision objectively assessed by SPVERA with the illiterate E Acuity in normal subjects. METHODS: The effect of artificially induced image defocusing on visual acuities (Vas) measured by SPVEP and illiterate E chart measurement was studied in 14 normal subjects. SPVEP was recorded using vertical gratings of 8 different spatial frequencies ranging from 0.235 9 to 30.203 2 c/deg. The responses were averaged and displayed through the discrete Fourier transform on the monitor display. SPVERA was determined by extrapolating the PVER amplitude -spatial frequency function to baseline. SPVERA was compared with illiterate E visual acuity measured under the same conditions of optical defocus. RESULTS: With moderate defocusing (<+1.0 diopter (D),VA > 0.42), the SPVERA were equal to or poorer than the illiterate E visual acuity. With more defocus (> +2.0 diopter (D), VA < 0.3),the SPVERA became better than the illiterate E chart Visual Acuity. The correlation between the two acuities was r2 = 0.919. CONCLUSION: SPVERA and the illiterate E visual acuity correlated to a certain degree. We conclude that the SPVERA is effective in estimating vision objectively.

[Visual electrophysiology and objective visual function].

This review summarizes objective assessment of visual function using visual electrophysiology. Objective assessment of visual acuity using pattern visual evoked potential (PVEP) and sweep pattern visual evoked potential (SPVEP), objective assessment of visual field using multifocal visual evoked potential (MVEP).