Low-Power Artificial Neural Network Perceptron Based on Monolayer MoS2.

Machine learning and signal processing on the edge are poised to influence our everyday lives with devices that will learn and infer from data generated by smart sensors and other devices for the Internet of Things. The next leap toward ubiquitous electronics requires increased energy efficiency of processors for specialized data-driven applications. Here, we show how an in-memory processor fabricated using a two-dimensional materials platform can potentially outperform its silicon counterparts in both standard and nontraditional Von Neumann architectures for artificial neural networks. We have fabricated a flash memory array with a two-dimensional channel using wafer-scale MoS2. Simulations and experiments show that the device can be scaled down to sub-micrometer channel length without any significant impact on its memory performance and that in simulation a reasonable memory window still exists at sub-50 nm channel lengths. Each device conductance in our circuit can be tuned with a 4-bit precision by closed-loop programming. Using our physical circuit, we demonstrate seven-segment digit display classification with a 91.5% accuracy with training performed ex situ and transferred from a host. Further simulations project that at a system level, the large memory arrays can perform AlexNet classification with an upper limit of 50 000 TOpS/W, potentially outperforming neural network integrated circuits based on double-poly CMOS technology.

Ultrasound biomicroscopic analysis of iris-fixed acrylic intraocular lens in the absence of capsule support.

The aim of the study was to investigate postoperative complications and to determine with ultrasound biomicroscopy (UBM) the position of a foldable acrylic intraocular lens (IOL) implanted with a surgical technique of iris suturing in eyes without capsule support. Six eyes with iris-sutured IOLs were examined postoperatively with UBM. The examination included the position of haptics and their relationship to the surrounding structures. Other parameters studied were central anterior chamber depth, IOL iris contact and pigment dispersion. The position of all 12 haptics was determined. All haptics were in touch with the iris. Pigment dispersion was seen in two cases. The anterior chamber depth of the sutured eyes was normal. There was neither cystoid macular edema nor chronic uveitis postoperatively. In conclusion, in most cases surgical placement of iris-fixed lenses is a blind procedure and UBM could be an appropriate method to determine the position of IOL haptics postoperatively and to demonstrate that this surgical technique is reproducible, safe, and effective with the limitation of the difficulty of iris suturing IOL.

The role of ultrasound biomicroscopy in the diagnosis of temporal arteritis.

Temporal artery biopsy is considered the gold standard for the diagnosis of temporal arteritis (TA). The aim of the study was to determine whether ultrasound biomicroscopy (UBM) findings are useful in predicting the result (positive or negative) of temporal artery biopsy in patients with TA. Sixteen consecutive patients (ten women and six men) with clinical diagnosis of TA seen at the Ophthalmology Unit, S. Marta Hospital, Catania University, were included in the study. All patients were submitted to UBM (50-MHz transducer, P45 Plus Paradigm Medical Instruments, Salt Lake City, UT, USA) before temporal artery biopsy. The results were correlated with histopathologic changes. Seven patients presented histopathologic findings consistent with the diagnosis of TA. Thus, UBM findings of these patients were compared with those of the nine patients with negative biopsy. On UBM, we searched for the presence of a hypoechoic effect surrounding the walls of temporal arteries, so-called halo sign, as well as for the intra-arterial middle reflexive filling, so-called intra-arterial filling. The halo sign and intra-arterial filling were found in all seven (100%) patients with biopsy-proven TA. However, in five (55.5%) patients, the absence of these two parameters on UBM of patients with suspected TA strongly suggested that temporal artery biopsy would be negative (negative predictive value, 100%). On the other hand, four (44.5%) of nine patients with negative biopsy presented one of these two UBM findings. In conclusion, this preliminary work suggested UBM to be an appropriate noninvasive tool for morphological imaging and evaluation of temporal arteries, helpful in obtaining an indication of the side and segment for biopsy, and may play a role in predicting negative result of temporal artery biopsy in patients with TA; however, for the time being we cannot recommend any change in the current practice.

Severe proliferative diabetic retinopathy treated with vitrectomy or panretinal photocoagulation: a monocenter randomized controlled clinical trial.

OBJECTIVE: Visual and anatomic results of pars plana vitrectomy were compared with panretinal photocoagulation in cases of severe proliferative diabetic retinopathy. DESIGN: Monocenter randomized controlled clinical trial; randomization and allocation to trial group were carried out by a central computer system. PARTICIPANTS: We studied 180 eyes of 180 patients with severe proliferative diabetic retinopathy, half with tractional retinal detachment not involving the macula (n = 90). Some were treated by pars plana vitrectomy (n = 90) and some by panretinal photocoagulation (n = 90). METHODS: Eyes were randomly divided into 2 groups. Group 1 was treated with pars plana vitrectomy, membrane and internal limited membrane peeling, panretinal photocoagulation, and focal or grid macular laser. Group 2 was treated with panretinal photocoagulation and focal or grid macular laser. Follow-up was at least 12 months. RESULTS: Mean preoperative visual acuities and ophthalmic fundus characteristics were compared across groups. A year postoperation, visual acuity (the primary endpoint) in group 1 improved in 30 eyes (33%); was unchanged in 22 eyes (25%); and decreased in 38 eyes (42%). In group 2, visual acuity improved in 32 eyes (35%); was unchanged in 42 eyes (47%); and decreased in 16 eyes (18%). The percentage of improvement in the 2 groups was not statistically significant (p = 0.75), whereas the stabilized (p = 0.002) and worsened eyes (p = 0.0003) in group 1 and group 2 were significantly different. CONCLUSION: In cases of severe proliferative diabetic retinopathy, even with tractional retinal detachment not involving macula, it is best to defer vitrectomy unless definite progression that threatens the vascular center is documented.

Heavy versus standard silicone oil in the management of retinal detachment with macular hole in myopic eyes.

PURPOSE: The purpose of this study was to compare pars plana vitrectomy (PPV) with 1000 cSt silicone oil endotamponade and PPV with densiron endotamponade for retinal detachment with macular hole and posterior staphyloma in highly myopic eyes. PATIENTS AND METHODS: In a prospective study, 30 eyes of 30 patients were randomly assigned to PPV and densiron (n = 15) or PPV with silicone oil (n = 15). All eyes had laser photocoagulation of the macular hole rim after PPV. Silicone oil or densiron was removed 12 weeks after surgery. Patients were followed-up for 6 months after oil removal. RESULTS: In the densiron group, the retinal reattachment rate was 100% with densiron in situ and 87% after its removal, and in the silicone oil group, the retinal reattachment rate was 67% with silicone oil in situ and 53.4% after oil removal. Thus, PPV with densiron had a better anatomical success rate than silicone oil (P = 0.04 with endotamponade and P = 0.05 after endotamponade removal). In both groups, paired comparison of preoperative and postoperative best-corrected visual acuity was not statistically significant (P = 0.08). CONCLUSION: Pars plana vitrectomy with densiron is a preferred surgical procedure for the repair of macular hole retinal detachment in highly myopic eyes with posterior staphyloma.

Transconjunctival sutureless 25-gauge versus 20-gauge standard vitrectomy: correlation between corneal topography and ultrasound biomicroscopy measurements of sclerotomy sites.

PURPOSE: To determine the correlation between corneal shape changes and ultrasound biomicroscopy (UBM) findings at the sclerotomy sites in conventional 20-gauge (G) pars plana vitrectomy (PPV) and 25-G transconjunctival sutureless vitrectomy (TSV) and to compare the effectiveness of the two surgical methods. DESIGN: Prospective, comparative, observational case series. METHODS: Sixty consecutive eyes (60 patients) undergoing primary 3-port PPV. Thirty eyes (30 patients, group 20-G) were treated with 20-G standard PPV and 30 eyes (30 patients, group 25-G) with 25-G TSV. We compared healing of the sclerotomy sites in the two groups. We determined the correlation between corneal shape changes (surgically induced astigmatism) measured by video keratography and the durations of scleral healing cicatrization by UBM within each group. RESULTS: UBM examination showed that the 20-G sclerotomy sites took about 8 weeks to heal, measured as complete opposition, whereas healing of the 25-G TSV sclerotomy was quite rapid, with complete scleral opposition in about 4 weeks. Corneal topography analysis showed, during the early postoperative period, a surgically induced steepening of the cornea in both groups (20 G, 3.08 +/- 0.56 diopters and 25 G, 0.805 +/- 0.61 diopters, P < 0.001, Mann-Whitney test), which then decreased gradually, recovering to the preoperative level within two months in group 20 G (P > 0.05) and 1 month in group 25 G (P > 0.05). We found a strong statistical correlation between the mean surgically induced keratometric astigmatism and the mean UBM measures of scleral healing (r = 0.99 for group 20 G and r = 0.97 for group 25 G). CONCLUSION: After PPV, astigmatic changes are especially significant in the early postoperative period in 20-G group; the 25-G TSV system results in faster reduction of surgically induced keratometric astigmatism because of rapid cicatrization of the sclerotomy sites.

Keratoconus staging: a computer-assisted ultrabiomicroscopic method compared with videokeratographic analysis.

PURPOSE: The aim of this study was to introduce a new paradigm for keratoconus assessment, the keratoconus index (KI), generated from the ratio of peripheral corneal thickness (PCT) to the thinnest corneal thickness (TCT), and calculated by a computer-assisted procedure after ultrabiomicroscope (UBM) examination. Then we compared KI and the keratoconus severity index (KSI), obtained by videokeratography in patients with different stages of keratoconus. METHODS: We studied 60 eyes with different forms of keratoconus using the TMS-3 autotopographer, provided with a keratoconus screening program (using Smolek-Klyce methods) and the commercial version of the ultrasound biomicroscope (Paradigm UBM Plus Model P45) equipped with a 50-MHz probe, which was provided with our computer-assisted program. The proportion test Z and the correlation coefficient R were applied to the outcomes. RESULTS: The keratoconus severity index, KSI, obtained by color-coded videokeratographic maps, was in the range 95% to 32% (mean 52.22%). By means of UBM examination, we obtained 60 images and found values of TCT 0.278-0.592 mm and PCT 0.475-0.704 mm. Applying the computer-assisted method, we obtained values for KI of 1.112-2.159 (mean 1.428). CONCLUSIONS: KI is correlated as well as KSI with the severity of the keratoconus (R = 0.76, P < 0.0001). It can be used as a similar parameter to measure the evolution of the disease, on the basis of corneal thickness rather than the curvature.