Application of convolutional neural networks for distal radio-ulnar fracture detection on plain radiographs in the emergency room

Objective@#Recent studies have suggested that deep-learning models can satisfactorily assist in fracture diagnosis. We aimed to evaluate the performance of two of such models in wrist fracture detection. @*Methods@#We collected image data of patients who visited with wrist trauma at the emergency department. A dataset extracted from January 2018 to May 2020 was split into training (90%) and test (10%) datasets, and two types of convolutional neural networks (i.e., DenseNet-161 and ResNet-152) were trained to detect wrist fractures. Gradient-weighted class activation mapping was used to highlight the regions of radiograph scans that contributed to the decision of the model. Performance of the convolutional neural network models was evaluated using the area under the receiver operating characteristic curve. @*Results@#For model training, we used 4,551 radiographs from 798 patients and 4,443 radiographs from 1,481 patients with and without fractures, respectively. The remaining 10% (300 radiographs from 100 patients with fractures and 690 radiographs from 230 patients without fractures) was used as a test dataset. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of DenseNet-161 and ResNet-152 in the test dataset were 90.3%, 90.3%, 80.3%, 95.6%, and 90.3% and 88.6%, 88.4%, 76.9%, 94.7%, and 88.5%, respectively. The area under the receiver operating characteristic curves of DenseNet-161 and ResNet-152 for wrist fracture detection were 0.962 and 0.947, respectively. @*Conclusion@#We demonstrated that DenseNet-161 and ResNet-152 models could help detect wrist fractures in the emergency room with satisfactory performance.

Application of convolutional neural networks for distal radio-ulnar fracture detection on plain radiographs in the emergency room

Objective@#Recent studies have suggested that deep-learning models can satisfactorily assist in fracture diagnosis. We aimed to evaluate the performance of two of such models in wrist fracture detection. @*Methods@#We collected image data of patients who visited with wrist trauma at the emergency department. A dataset extracted from January 2018 to May 2020 was split into training (90%) and test (10%) datasets, and two types of convolutional neural networks (i.e., DenseNet-161 and ResNet-152) were trained to detect wrist fractures. Gradient-weighted class activation mapping was used to highlight the regions of radiograph scans that contributed to the decision of the model. Performance of the convolutional neural network models was evaluated using the area under the receiver operating characteristic curve. @*Results@#For model training, we used 4,551 radiographs from 798 patients and 4,443 radiographs from 1,481 patients with and without fractures, respectively. The remaining 10% (300 radiographs from 100 patients with fractures and 690 radiographs from 230 patients without fractures) was used as a test dataset. The sensitivity, specificity, positive predictive value, negative predictive value, and accuracy of DenseNet-161 and ResNet-152 in the test dataset were 90.3%, 90.3%, 80.3%, 95.6%, and 90.3% and 88.6%, 88.4%, 76.9%, 94.7%, and 88.5%, respectively. The area under the receiver operating characteristic curves of DenseNet-161 and ResNet-152 for wrist fracture detection were 0.962 and 0.947, respectively. @*Conclusion@#We demonstrated that DenseNet-161 and ResNet-152 models could help detect wrist fractures in the emergency room with satisfactory performance.

Prediction of Suicidal Ideation among Korean AdultsUsing Machine Learning: A Cross-Sectional Study

Objective@#Suicidal ideation (SI) precedes actual suicidal event. Thus, it is important for the prevention of suicide to screen the individualswith SI. This study aimed to identify the factors associated with SI and to build prediction models in Korean adults using machinelearning methods. @*Methods@#The 2010–2013 dataset of the Korea National Health and Nutritional Examination Survey was used as the training dataset(n=16,437), and the subset collected in 2015 was used as the testing dataset (n=3,788). Various machine learning algorithms were appliedand compared to the conventional logistic regression (LR)-based model. @*Results@#Common risk factors for SI included stress awareness, experience of continuous depressive mood, EQ-5D score, depressivedisorder, household income, educational status, alcohol abuse, and unmet medical service needs. The prediction performances of themachine learning models, as measured by the area under receiver-operating curve, ranged from 0.794 to 0.877, some of which were betterthan that of the conventional LR model (0.867). The Bayesian network, LogitBoost with LR, and ANN models outperformed the conventionalLR model. @*Conclusion@#A machine learning-based approach could provide better SI prediction performance compared to a conventional LRbasedmodel. These may help primary care physicians to identify patients at risk of SI and will facilitate the early prevention of suicide.Psychiatry Investig 2020;17(4):331-340

Clinical Features and Surgical Outcomes of Primary Rhegmatogenous Retinal Detachment according to Age

PURPOSE: To investigate the clinical features and surgical outcomes of rhegmatogenous retinal detachment (RRD) requiring surgery according to age. METHODS: Medical records of patients who underwent surgery for primary RRD between January 2008 and March 2016 were reviewed retrospectively. Patients were classified into two groups according to age at diagnosis: the under-40 group and the over-40 group. The two groups were compared in terms of demographic features, ocular manifestation, operating methods, primary anatomical success rate, and visual outcome. RESULTS: One hundred and forty-four eyes from 144 patients were included. Mean subject age was 48.6 ± 16.9 years old. The under-40 group involved 42 eyes from 42 patients, and the over-40 group included 102 eyes from 102 patients. Symptom duration was shorter in the under-40 group compared to the over-40 group (7.6 ± 10.7 days vs. 14.5 ± 24.4 days; p = 0.029). Proliferative vitreoretinopathy (PVR) occurred more frequently in the under-40 group (40.0% vs. 17.4%, p = 0.007) than in the over-40 group. The anatomical success rate of primary surgery was significantly different between the two groups; 78.6% in the under-40 group and 91.2% in the over-40 group (p = 0.038). Preoperative PVR increased the rate of anatomical failure (40.0% vs. 6.2%, p < 0.001). The visual outcomes were not significantly different between the two groups. CONCLUSIONS: RRD is combined with PVR more frequently in young patients than in old patients, which increases the failure rate of primary re-attachment surgery.

Intravitreal Injection of Dexamethasone Implant during Cataract Surgery in Patients with Noninfectious Uveitis

PURPOSE: To investigate the efficacy and safety of intravitreal dexamethasone implant for controlling postoperative inflammation among uveitis patients undergoing cataract extraction. METHODS: Ten eyes with noninfectious uveitis underwent phacoemulsification with intraocular lens implantation followed by intravitreal injection of 0.7-mg dexamethasone implant (implant group) between February 2011 and January 2014. Twenty age- and gender-matched controls who received cataract surgery without implantation during the same period were recruited (non-implant group). Medical records of the subjects were retrospectively reviewed and 6-month postoperative clinical outcomes were compared between the 2 groups. RESULTS: The mean age was 42.30 +/- 15.81 years in the implant group and 45.65 +/- 13.63 years in the non-implant group. The 2 groups were similar in terms of age, gender, preoperative inflammatory status, and preoperative visual acuity (p = 0.552, 1.000, 0.133 and 0.767, respectively). After surgery, oral steroid was used in the non-implant group (8.8 +/- 1.5 mg/day on average) and the implant group (3.5 +/- 1.3 mg/day; p = 0.029). Visual acuity (log MAR) improved significantly in both groups (p = 0.789) with no significant difference between the 2 groups. Postoperative recurrence rates of uveitis were reduced more (40%) in the implant group than in the non-implant group (50%) but without significance (p = 0.709). Elevated intraocular pressure > or =25 mm Hg occurred in 3 eyes (30%) in the implant group and 4 eyes (20%) in the non-implant group (p = 0.657), of which 1 in each group required a filtering surgery. Otherwise, no significant complications developed in either group. CONCLUSIONS: Intravitreal dexamethasone implants help reduce conventional oral steroid dosage for controlling postoperative inflammation. Dexamethasone implants could be an effective and safe alternative to control the inflammation after cataract surgery in uveitis patients.

Horner's Syndrome and Contralateral Abducens Nerve Palsy Associated with Zoster Meningitis

A 55-year-old woman presented with diplopia following painful skin eruptions on the right upper extremity. On presentation, she was found to have 35 prism diopters of esotropia and an abduction limitation in the left eye. Two weeks later, she developed blepharoptosis and anisocoria with a smaller pupil in the right eye, which increased in the darkness. Cerebrospinal fluid analysis showed pleocytosis and a positive result for immunoglobulin G antibody to varicella zoster virus. She was diagnosed to have zoster meningitis with Horner's syndrome and contralateral abducens nerve palsy. After intravenous antiviral and steroid treatments, the vesicular eruptions and abducens nerve palsy improved. Horner's syndrome and diplopia resolved after six months. Here we present the first report of Horner's syndrome and contralateral abducens nerve palsy associated with zoster meningitis.

Short-term Clinical Outcome of Patterned Scanning Laser Photocoagulation With Short Exposure Time in Diabetic Retinopathy

PURPOSE: To investigate the clinical effect and complications of patterned scanning laser photocoagulation with short exposure time in diabetic retinopathy. METHODS: A prospective study was performed on patients with diabetic retinopathy who required panretinal photocoagulation (PRP). Twenty-nine eyes of twenty five consecutive patients underwent patterned scanning laser photocoagulation with exposure time of 0.02 sec over the entire retina. Pain score at treatment, best-corrected visual acuity, the development of macular edema, regression of neovascularization in fluorescein angiography, and related complications were monitored during a three-month follow-up. RESULTS: Treatment time for PRP was much reduced to 6.1 min on the entire retina. Pain during the photocoagulation was moderate, and enabled patients to finish PRP with ease. During a three-month follow-up, visual acuity was well preserved. Foveal minimum thickness increased at 1 month after treatment and maintained in a study period of three months. Macular volume showed significant increase (3.6%) at 1 month after treatment, maximum increase (9.8%) at 2 months, and persistent increase (4.9%) at 3 months. In every patient with very severe nonproliferative diabetic retinopathy, retina was maintained, and in 10 of 18 eyes with proliferative diabetic retinopathy, neovascularization regressed during follow-up period. No significant complication occurred except a transient vitreous hemorrhage in one eye. CONCLUSIONS: Patterned scanning laser photocoagulation with short exposure time may induce transient macular edema, but the visual acuity was preserved without any other significant complication. It is considered to be a safe and efficient treatment method in diabetic retinopathy.

Two Cases of Cosmetic Treatment of Limbal Dermoid

PURPOSE: To report two cases of cosmetic treatment of limbal dermoid, which consist of local excision of the limbal dermoid, corneal tattooing, and amniotic membrane transplantation. METHODS: Dermoid excision, corneal tattooing, and amniotic membrane transplantation were carried out in both a girl and a woman who each had a limbal dermoid in their left eyes. Postoperatively, they were followed up for more than six months. RESULTS: No specific complications occurred. The patients were satisfied with the surgical results. After a few months the conjunctiva was found to grow into the tattooed cornea. CONCLUSIONS: In the treatment of limbal dermoid, local excision combined with corneal tattooing and amniotic membrane transplantation can be a cosmetically successful alternative choice.

Cosmetic Repair of Nevus of Ota

PURPOSE: To report a case of cosmetically successful treatment of nevus of Ota performed by scleral painting and scleral allograft transplantation. METHODS: Scleral painting and allograft transplantation were done in a 28-year-old woman who had cosmetic problem due to nevus of Ota, and she was followed up for 5 months postoperatively. RESULTS: The patient had no specific postoperative complication and was satisfied with the surgical results. CONCLUSIONS: For patients of nevus of Ota who have not been treated successfully, scleral painting and allograft transplation can be an effective cosmetic treatment.