Graham Mann

Detection of glaucoma progression is crucial to managing patients, permitting individualized care plans and treatment. It is a challenging task requiring the assessment of structural changes to the optic nerve head and functional changes based on visual field testing. Artificial intelligence, especially deep learning techniques, has shown promising results in many applications, including glaucoma diagnosis. This paper proposes a two-stage computational learning pipeline for detecting glaucoma progression using only fundus photographs. In the first stage, a deep learning model takes a time series of fundus photographs as input and outputs a vector of predictions where each element represents the overall rate of change in visual field (VF) sensitivity values for a sector (region) of the optic nerve head (ONH). We implemented two deep learning models—ResNet50 and InceptionResNetV2—for this stage. In the second stage, a binary classifier (weighted logistic regression) takes the predicted vector as input to detect progression. We also propose a novel method for constructing annotated datasets from temporal sequences of clinical fundus photographs and corresponding VF data suitable for machine learning. Each dataset element comprises a temporal sequence of photographs together with a vector-valued label. The label is derived by computing the pointwise linear regression of VF sensitivity values at each VF test location, mapping these locations to eight ONH sectors, and assigning the overall rate of change in each sector to one of the elements of the vector. We used a retrospective clinical dataset with 82 patients collected at multiple timepoints over five years in our experiments. The InceptionResNetV2-based implementation yielded the best performance, achieving detection accuracies of 97.28 ± 1.10 % for unseen test data (i.e., each dataset element is unseen but originates from the same set of patients appearing in the training dataset), and 87.50 ± 0.70 % for test data from unseen patients (training and testing patients are entirely different). The testing throughput was 11.60 ms per patient. These results demonstrate the efficacy of the proposed method for detecting glaucoma progression from fundus photographs.

Optic disc (OD) segmentation from retinal images is crucial for diagnosing, assessing, and tracking the progression of several sight-threatening diseases. This paper presents a deep machine-learning method for semantically segmenting OD from retinal images. The method is named multiscale guided attention network (MSGANet-OD), comprising encoders for extracting multiscale features and decoders for constructing segmentation maps from the extracted features. The decoder also includes a guided attention module that incorporates features related to structural, contextual, and illumination information to segment OD. A custom loss function is proposed to retain the optic disc's geometrical shape (i.e., elliptical) constraint and to alleviate the blood vessels' influence in the overlapping region between the OD and vessels. MSGANet-OD was trained and tested on an in-house clinical color retinal image dataset captured during ophthalmodynamometry as well as on several publicly available color fundus image datasets, e.g., DRISHTI-GS, RIM-ONE-r3, and REFUGE1. Experimental results show that MSGANet-OD achieved superior OD segmentation performance from ophthalmodynamometry images compared to widely used segmentation methods. Our method also achieved competitive results compared to state-of-the-art OD segmentation methods on public datasets. The proposed method can be used in automated systems to quantitatively assess optic nerve head abnormalities (e.g., glaucoma, optic disc neuropathy) and vascular changes in the OD region.

Intracranial pressure (ICP) measurement is vital in managing and treating patients with brain injuries, brain tumors, and other neuro-ophthalmic conditions. Invasive methods have reached a high standard of accuracy but pose multiple risks to patients and require specialized resources. In contrast, non-invasive methods indirectly estimate ICP based on related physiological properties. This paper surveys imaging-based methods that utilize cerebro-ophthalmic anatomy and fluid dynamics. The theoretical basis of each method is presented, including the physiological attributes and imaging modality used. Details of empirical studies are also presented, including subjects (type, population, medical indication, captured anatomies), methodology, and evaluations of ICP estimation accuracy. The pros and cons of each method are presented in terms of complexity, patient acceptability, operating expertise, potential for clinical application, costs, and time. Whilst none of the methods in the survey has to date achieved the reliability and accuracy of invasive methods used clinically, those based on transcranial and orbital ultrasonography and direct retinal imaging demonstrate great promise. Refinements to these methods, including the application of modern deep learning techniques, offer the prospect of non-invasive, clinically acceptable, accurate and automatic measurement of ICP.

Background

Retinal and optic disc images are used to assess changes in the retinal vasculature. These can be changes associated with diseases such as diabetic retinopathy and glaucoma or induced using ophthalmodynamometry to measure arterial and venous pressure. Key steps toward automating the assessment of these changes are the segmentation and classification of the veins and arteries. However, such segmentation and classification are still required to be manually labelled by experts. Such automated labelling is challenging because of the complex morphology, anatomical variations, alterations due to disease and scarcity of labelled data for algorithm development. We present a deep machine learning solution called the multiscale guided attention network for retinal artery and vein segmentation and classification (MSGANet-RAV).

Methods

MSGANet-RAV was developed and tested on 383 colour clinical optic disc images from LEI-CENTRAL, constructed in-house and 40 colour fundus images from the AV-DRIVE public dataset. The datasets have a mean optic disc occupancy per image of 60.6% and 2.18%, respectively. MSGANet-RAV is a U-shaped encoder-decoder network, where the encoder extracts multiscale features, and the decoder includes a sequence of self-attention modules. The self-attention modules explore, guide and incorporate vessel-specific structural and contextual feature information to segment and classify central optic disc and retinal vessel pixels.

Results

MSGANet-RAV achieved a pixel classification accuracy of 93.15%, sensitivity of 92.19%, and specificity of 94.13% on LEI-CENTRAL, outperforming several reference models. It similarly performed highly on AV-DRIVE with an accuracy, sensitivity and specificity of 95.48%, 93.59% and 97.27%, respectively.

Conclusion

The results show the efficacy of MSGANet-RAV for identifying central optic disc and retinal arteries and veins. The method can be used in automated systems designed to assess vascular changes in retinal and optic disc images quantitatively.

The need for software applications that can assist with mental disorders has never been greater. Individuals suffering from mental illnesses often avoid consultation with a psychotherapist, because they do not realize the need, or because they cannot or will not face the social and economic consequences, which can be severe. Between ideal treatment by a human therapist and self-help websites lies the possibility of a helpful interaction with a language-using computer. A model of empathic response planning for sentence generation in a forthcoming automated psychotherapist is described here. The model combines emotional state tracking, contextual information from the patient’s history and continuously updated therapeutic goals to form suitable conceptual graphs that may then be realized as suitable textual sentences.

This paper proposes an architecture that explores a gap in the spectrum of existing strategies for robot control mode switching in adjustable autonomy. In situations where the environment is reasonably known and/or predictable, pre-planning these control changes could relieve robot operators of the additional task of deciding when and how to switch. Such a strategy provides a clear division of labour between the automation and the human operator(s) before the job even begins, allowing for individual responsibilities to be known ahead of time, limiting confusion and allowing rest breaks to be planned. Assigned Responsibility is a new form of adjustable autonomy-based teleoperation that allows the selective inclusion of automated control elements at key stages of a robot operation plan’s execution. Progression through these stages is controlled by automatic goal accomplishment tracking. An implementation is evaluated through engineering tests and a usability study, demonstrating the viability of this approach and offering insight into its potential applications.

Reports on the field testing of robots technology. The trouble with field-testing robots is that we are taking complex machines out of the laboratory and into the dirt: natural, unstructured environments that cannot be easily characterized or measured. There they could be doing imperfectly characterized tasks. We expect robots to be behaviorally flexible so describing a typical task will generally underspecify actual usage. The machine design, task, and environment are not orthogonal factors either, since they might interact in complicated ways. As if all this was not enough, most field robots are still teleoperated, which adds the attendant problems of evaluating the human controller and interface. Published work in this area tends to focus on demonstrating the robot's fitness for purpose based on specific requirements, often according to he contingencies of practical funding. Too often that commits the work to studies of performance on tasks that are not necessarily well understood, or even particularly well described, and to measurements within environments that cannot be duplicated.

Controlled testing on standard tasks and within standard environments can provide meaningful performance comparisons between robots of heterogeneous design. But because they must perform practical tasks in unstructured, and therefore non-standard, environments, the benefits of this approach have barely begun to accrue for field robots. This work describes a desert trial of six student prototypes of astronaut-support robots using a set of standardized engineering tests developed by the US National Institute of Standards and Technology (NIST), along with three operational tests in natural Mars-like terrain. The results suggest that standards developed for emergency response robots are also applicable to the astronaut support domain, yielding useful insights into the differences in capabilities between robots and real design improvements. The exercise shows the value of combining repeatable engineering tests with task-specific application-testing in the field.

As with other types of evidence, the courts make no presumption that digital evidence is reliable without some evidence of empirical testing in relation to the theories and techniques associated with its production. The issue of reliability means that courts pay close attention to the manner in which electronic evidence has been obtained and in particular the process in which the data is captured and stored. Previous process models have tended to focus on one particular area of digital forensic practice, such as law enforcement, and have not incorporated a formal description. We contend that this approach has prevented the establishment of generally accepted standards and processes that are urgently needed in the domain of digital forensics. This paper presents a generic process model as a step towards developing such a generally-accepted standard for a fundamental digital forensic activity-the acquisition of digital evidence.

Australia is an ideal testing ground in preparation for the robotic and human exploration of Mars. Numerous sites with landforms or processes analogous to those on Mars are present and the deserts of central Australia provide a range of locations for free-ranging Mars analogue mission simulations. The latest developments in testing technologies and strategies for exploration in Australian Mars analogues are reviewed. These include trials of analogue space suits based on mechanical counter pressure technology and the development of an analogue, crewed, pressurized rover for long-range exploration. Field science activities and instrumentation testing relevant to robotic and future crewed missions are discussed. Australian-led human factors research undertaken during expeditions to Mars analogue research stations and expeditions to Antarctica are also reviewed. Education and public outreach activities related to Mars analogue research in Australia are also detailed.