Tom H. Williamson

Automatic detection of diabetic retinopathy using an artificial neural network: a screening tool.

AIMS: To determine if neural networks can detect diabetic features in fundus images and compare the network against an ophthalmologist screening a set of fundus images. METHODS: 147 diabetic and 32 normal images were captured from a fundus camera, stored on computer, and analysed using a back propagation neural network. The network was trained to recognise features in the retinal image. The effects of digital filtering techniques and different network variables were assessed. 200 diabetic and 101 normal images were then randomised and used to evaluate the networks performance for the detection of diabetic retinopathy against an ophthalmologist. RESULTS: Detection rates for the recognition of vessels, exudates, and haemorrhages were 91.7%, 93.1%, and 73.8% respectively. When compared with the results of the ophthalmologist, the network achieved a sensitivity of 88.4% and a specificity of 83.5% for the detection of diabetic retinopathy. CONCLUSIONS: Detection of vessels, exudates, and haemorrhages was possible, with success rates dependent upon preprocessing and the number of images used in training. When compared with the ophthalmologist, the network achieved good accuracy for the detection of diabetic retinopathy. The system could be used as an aid to the screening of diabetic patients for retinopathy.

Color Doppler ultrasound imaging of theeye and orbit

Color Doppler imaging is a non-invasive ultrasound procedure which permits simultaneous gray scale imaging of structure and color-coded imaging of blood velocity. This improved technique allows the user to identify even very small blood vessels, such as those supplying the eye, from which measures of blood velocity and vascular resistance can be obtained. In the past five years, color Doppler imaging has found a number of applications in ophthalmology. A common examination procedure and expected normal values have been established, and the technique is becoming routinely employed to evaluate orbital vasculature in some medical centers. Color Doppler imaging has successfully demonstrated changes in orbital hemodynamics associated with a variety of pathological conditions, including central retinal artery and vein occlusions, cranial arteritis, nonarteritic ischemic optic neuropathy, and carotid disease. In addition, the method has been used to detect the vascularization of orbital and ocular tumors, as well as to investigate altered hemodynamics associated with diseases such as glaucoma and diabetic retinopathy.

Test/retest reproducibility of color Doppler imaging assessment of blood flow velocity in orbital vessels

PurposeThe purpose of this study was to evaluate the test-retest reproducibility of measures of blood flow velocities using color Doppler imaging (CDI) in orbital arteries. Patients and MethodsMeasures of peak systolic velocity, end-diastolic velocity, and resistive index were performed in a group of 15 normal tension glaucoma patients and in 15 healthy subjects using the Siemens Quantum 2000 CDI system with a 7.5 MHz linear probe. After each velocity measure in each vessel, the probe was removed and then replaced to repeat the measurement −2 min later. ResultsFor all subjects, the coefficients of reliability for measures of peak systolic velocity were 12% in the ophthalmic artery, 25% in the central retinal artery, and 19% in a short posterior ciliary artery. For end diastolic velocity, the coefficients were 6%, 11%, and 25%, respectively. The coefficients for resistive index were 4%, 11%, and 38%. ConclusionCDI produces highly reproducible measures in the ophthalmic artery. Measures in the central retinal artery are somewhat more variable but seem reasonably reproducible. Short posterior ciliary measurements were the most variable, suggesting that current methods for assessing these vessels may not be sufficiently reliable.

Ocular blood flow measurement.

Many techniques have been devised to measure the haemodynamics of the human and animal eye. In this perspective these are outlined and their use in ophthalmic investigation summarised. Some have exploited the ability of an observer to directly visualise the retinal vasculature by optical means, others have been designed to study the haemodynamics ofthe invisible parts of the eye such as the choroid, optic nerve head, and ciliary body. Although useful in ophthalmic investigation, none have satisfied all of the requirements of the researchers in this field and most have not achieved regular use in clinical practice. In any examination of blood flow a multitude of variables must be studied (Table 1). The interrelation ofthese variables must be determined while considering physical or physiological principles (Table 2) which are often not strictly applicable to the vasculature for example, the Hagen Poiseuille law was described for a rigid tube and not for elastic walled tubes such as blood vessels. In the human, study ofthe circulation is further hindered by the requirement for a noninvasive and safe method for obtaining measurements. Study of the haemodynamics must be performed if we are to understand the mechanisms leading to the large variety of vascular diseases which affect the eye. The blood flow to the eye is of particular interest because: (1) Many localised and systemic disorders affect the vasculature of the eye. (2) The eye has unusual haemodynamic properties because the tissues are subjected to a high intraocular pressure. (3) Ocular blood flow is autoregulated for example, during changes in retinal illumination, blood pressure, or posture. (4) Pharmacological agents which are routinely used in systemic and ocular diseases may affect the blood supply of the eye.

Functional imaging of the retina using the multifocal electroretinograph: a control study.

BACKGROUND: A new technique exists that enables functional mapping of the retina. A control population was examined to obtain normative values and to assess the reproducibility of this new test. METHODS: Twenty healthy volunteers were tested using a 61 hexagonal array stimulus with a 14 minute recording period. Median 5th and 95th percentiles were determined for implicit times and amplitude measures for the 61 test areas. Repeat measurements were performed on 10 individuals. Wilcoxon and Bland and Altman techniques were used to quantify reproducibility of the test. RESULTS: The implicit time of the wave-form components was not found to vary over the retina (peak or b-wave component, 35.52 (1.4) ms; trough or a-wave component, 17.76 (0.8) ms). Reproducibility was found to decrease with eccentricity (coefficient of repeatability 17.4% for the central area increasing to 30.3% for the peripheral ring). CONCLUSIONS: The findings suggest that reproducibility, although variable with eccentricity, is comparable with conventional electrophysiology. These limits of variation were used to assign confidence intervals to individual retinal areas and will be used (future work) in the examination of diseased states.

Colour Doppler velocimetry of the arterial vasculature of the optic nerve head and orbit

Blood flow in the orbital vasculature can be investigated by colour Doppler velocimetry which allows simultaneous imaging of B-scan and Doppler ultrasound. In this paper we review the results of our recent studies on the arterial vasculature of the orbit and optic nerve head. A control study is described in which reliable measurements were obtained from the ophthalmic artery at the orbital apex, the medial orbit, and the complex of the central retinal artery and posterior ciliary arteries. Patients with occlusion of the central retinal artery and posterior ciliary arteries were examined. The results indicate that readings at the optic nerve head primarily represent blood flow in the central retinal artery. A study of the effects of posture on velocimetry demonstrated no change in the measurements obtained. Finally, topical timolol 0.5% was found to reduce the resistive indices of flow in the central retinal artery.

Comparison of repeatability of the multifocal electroretinogram and Humphrey perimeter.

Functional mapping of the retina by multifocal electroretinographic recordings is now possible. We compared the normal range, repeatability and response topography of this new technique with conventional static Humphrey perimetry to assess its suitability in clinical practice. The multifocal technique was performed on 60 age-matched controls. Measures of repeatability and reproducibility were obtained. Results were then compared with those obtained from a customized perimetry test. In both tests the coefficients of repeatability were found to decrease with eccentricity. The inherent measurement variation between techniques was comparable. Overall system variation indicates that the technique could be a useful tool at the clinical level.

Colour Doppler ultrasound in the management of a case of cranial arteritis.

Colour Doppler ultrasound allows simultaneous B scan and Doppler imaging and can be employed to determine the velocity of blood flow in the vasculature of the eye and orbit. We describe a case of cranial arteritis (giant cell arteritis) in which serial velocimetry recordings were obtained. At one stage in the disease process no blood flow was detectable in the orbit despite previously reliable recordings. This coincided with a deterioration of the clinical state of the patient as signified by recurrent anterior ischaemic optic neuropathy despite controlled symptomatology and erythrocyte sedimentation rate by prednisolone therapy. Subsequent increase in the immunosuppressive therapy was accompanied by a return of blood flow in the orbit. Colour Doppler ultrasound may prove to be a useful examination technique in the diagnosis and management of cranial arteritis.

The effect of filter bandwidth on the multifocal electroretinogram

The effect of filter bandwith on the multifocal electroretinogram was assessed by means of theoretical calculation, electronic simulation and real multifocal electroretinogram recordings. Waveform distortion by high-pass filtering on simulated square waves, normal electroretinogram waveforms and negative electroretinogram waveforms was demonstrated. The theory of the effect of differentiation on electroretinographic waveform shape by electronic filtering indicates that little effect would be observed by changing the input filter cut-off for normal electroretinographic waveform shapes. However, negative electroretinogram waveforms are differentiated when the high-pass filter setting is increased. The differentiation effect artificially recreates a positive component that could be mistaken as a b-wave component. To eliminate this effect when recording multifocal electroretinograms, a high-pass filter setting of less than 1 Hz should be used to preserve the true electroretinographic waveform shape.

Co-existence of exfoliation syndrome, previous iris surgery, and heterochromia.

Abstract. A case is described where exfoliation syndrome developed in a relatively young patient with heterochromia. The patient had previously undergone large radial iridotomies as part of penetrating keratoplasty procedures. This case illustrates an association between iris surgery and early manifestation of exfoliation syndrome.