Eye tracking: empirical foundations for a minimal reporting guideline.

In this paper, we present a review of how the various aspects of any study using an eye tracker (such as the instrument, methodology, environment, participant, etc.) affect the quality of the recorded eye-tracking data and the obtained eye-movement and gaze measures. We take this review to represent the empirical foundation for reporting guidelines of any study involving an eye tracker. We compare this empirical foundation to five existing reporting guidelines and to a database of 207 published eye-tracking studies. We find that reporting guidelines vary substantially and do not match with actual reporting practices. We end by deriving a minimal, flexible reporting guideline based on empirical research (Section "An empirically based minimal reporting guideline").

Small eye movements cannot be reliably measured by video-based P-CR eye-trackers.

For evaluating whether an eye-tracker is suitable for measuring microsaccades, Poletti & Rucci (2016) propose that a measure called 'resolution' could be better than the more established root-mean-square of the sample-to-sample distances (RMS-S2S). Many open questions exist around the resolution measure, however. Resolution needs to be calculated using data from an artificial eye that can be turned in very small steps. Furthermore, resolution has an unclear and uninvestigated relationship to the RMS-S2S and STD (standard deviation) measures of precision (Holmqvist & Andersson, 2017, p. 159-190), and there is another metric by the same name (Clarke, Ditterich, Drüen, Schönfeld, and Steineke 2002), which instead quantifies the errors of amplitude measurements. In this paper, we present a mechanism, the Stepperbox, for rotating artificial eyes in arbitrary angles from 1' (arcmin) and upward. We then use the Stepperbox to find the minimum reliably detectable rotations in 11 video-based eye-trackers (VOGs) and the Dual Purkinje Imaging (DPI) tracker. We find that resolution correlates significantly with RMS-S2S and, to a lesser extent, with STD. In addition, we find that although most eye-trackers can detect some small rotations of an artificial eye, the rotations of amplitudes up to 2∘ are frequently erroneously measured by video-based eye-trackers. We show evidence that the corneal reflection (CR) feature of these eye-trackers is a major cause of erroneous measurements of small rotations of artificial eyes. Our data strengthen the existing body of evidence that video-based eye-trackers produce errors that may require that we reconsider some results from research on reading, microsaccades, and vergence, where the amplitude of small eye movements have been measured with past or current video-based eye-trackers. In contrast, the DPI reports correct rotation amplitudes down to 1'.

Eye-tracking data quality as affected by ethnicity and experimental design.

Lack of accuracy in eye-tracking data can be critical. If the point of gaze is not recorded accurately and reliably, the information obtained or action executed might be different from what the user intended. This study reports trackability, accuracy, and precision as indicators of eye-tracking data quality as measured at various head positions and light conditions for a sample of participants from three different ethnic groups. It was found that accuracy and precision for Asian participants was worse than that for African and Caucasian participants. No significant differences were found between the latter two ethnic groups. Operating distance had the largest effect on data quality, since it affected all indicators for all ethnic groups. Illumination had no significant effect on accuracy or precision, but the accuracy achieved by African and Caucasian participants was better when the stimulus was presented on a dark background. Large gaze angles proved to be detrimental for trackability for African participants, while accuracy and precision were also affected adversely by larger gaze angles for two of the ethnicities.

COVID-19 diagnosis using deep learning neural networks applied to CT images.

COVID-19, a deadly and highly contagious virus, caused the deaths of millions of individuals around the world. Early detection of the virus can reduce the virus transmission and fatality rate. Many deep learning (DL) based COVID-19 detection methods have been proposed, but most are trained on either small, incomplete, noisy, or imbalanced datasets. Many are also trained on a small number of COVID-19 samples. This study tackles these concerns by introducing DL-based solutions for COVID-19 diagnosis using computerized tomography (CT) images and 12 cutting-edge DL pre-trained models with acceptable Top-1 accuracy. All the models are trained on 9,000 COVID-19 samples and 5,000 normal images, which is higher than the COVID-19 images used in most studies. In addition, while most of the research used X-ray images for training, this study used CT images. CT scans capture blood arteries, bones, and soft tissues more effectively than X-Ray. The proposed techniques were evaluated, and the results show that NASNetLarge produced the best classification accuracy, followed by InceptionResNetV2 and DenseNet169. The three models achieved an accuracy of 99.86, 99.79, and 99.71%, respectively. Moreover, DenseNet121 and VGG16 achieved the best sensitivity, while InceptionV3 and InceptionResNetV2 achieved the best specificity. DenseNet121 and VGG16 attained a sensitivity of 99.94%, while InceptionV3 and InceptionResNetV2 achieved a specificity of 100%. The models are compared to those designed in three existing studies, and they produce better results. The results show that deep neural networks have the potential for computer-assisted COVID-19 diagnosis. We hope this study will be valuable in improving the decisions and accuracy of medical practitioners when diagnosing COVID-19. This study will assist future researchers in minimizing the repetition of analysis and identifying the ideal network for their tasks.

Convolutional Neural Network-Based Technique for Gaze Estimation on Mobile Devices.

Eye tracking is becoming a very popular, useful, and important technology. Many eye tracking technologies are currently expensive and only available to large corporations. Some of them necessitate explicit personal calibration, which makes them unsuitable for use in real-world or uncontrolled environments. Explicit personal calibration can also be cumbersome and degrades the user experience. To address these issues, this study proposes a Convolutional Neural Network (CNN) based calibration-free technique for improved gaze estimation in unconstrained environments. The proposed technique consists of two components, namely a face component and a 39-point facial landmark component. The face component is used to extract the gaze estimation features from the eyes, while the 39-point facial landmark component is used to encode the shape and location of the eyes (within the face) into the network. Adding this information can make the network learn free-head and eye movements. Another CNN model was designed in this study primarily for the sake of comparison. The CNN model accepts only the face images as input. Different experiments were performed, and the experimental result reveals that the proposed technique outperforms the second model. Fine-tuning was also performed using the VGG16 pre-trained model. Experimental results show that the fine-tuned results of the proposed technique perform better than the fine-tuned results of the second model. Overall, the results show that 39-point facial landmarks can be used to improve the performance of CNN-based gaze estimation models.

A cost function to determine the optimum filter and parameters for stabilising gaze data.

Prior to delivery of data, eye tracker software may apply filtering to correct for noise. Although filtering produces much better precision of data, it may add to the time it takes for the reporting of gaze data to stabilise after a saccade due to the usage of a sliding window. The effect of various filters and parameter settings on accuracy, precision and filter related latency is examined. A cost function can be used to obtain the optimal parameters (filter, length of window, metric and threshold for removal of samples and removal percentage). It was found that for any of the FIR filters, the standard deviation of samples can be used to remove 95% of samples in the window so than an optimum combination of filter related latency and precision can be obtained. It was also confirmed that for unfiltered data, the shape of noise, signified by RMS/STD, is around 2 as expected for white noise, whereas lower RMS/STD values were observed for all filters.

Visualization and quantification of eye tracking data for the evaluation of oculomotor function.

Oculomotor dysfunction may originate from physical, physiological or psychological causes and may be a marker for schizophrenia or other disorders. Observational tests for oculomotor dysfunction are easy to administer, but are subjective and transient, and it is difficult to quantify deviations. To date, video-based eye tracking systems have not provided a contextual overview of gaze data that integrates the eye video recording with the stimulus and gaze data together with quantitative feedback of metrics in relation to typical values. A system was developed with an interactive timeline to allow the analyst to scroll through a recording frame-by-frame while comparing data from three different sources. The visual and integrated nature of the analysis allows localisation and quantification of saccadic under- and overshoots as well as determination of the frequency and amplitude of catch-up and anticipatory saccades. Clinicians will be able to apply their expertise to diagnose disorders based on abnormal patterns in the gaze plots. They can use the line charts to quantify deviations from benchmark values for reaction time, saccadic accuracy and smooth pursuit gain. A clinician can refer to the eye video at any time to confirm that observed deviations originated from gaze behaviour and not from systemic errors.

The Effect of Real-time Headbox Adjustments on Data Quality.

Following a patent owned by Tobii, the framerate of a CMOS camera can be increased by reducing the size of the recording window so that it fits the eyes with minimum room to spare. The position of the recording window can be dynamically adjusted within the camera sensor area to follow the eyes as the participant moves the head. Since only a portion of the camera sensor data is communicated to the computer and processed, much higher framerates can be achieved with the same CPU and camera. Eye trackers can be expected to present data at a high speed, with good accuracy and precision, small latency and with minimal loss of data while allowing participants to behave as normally as possible. In this study, the effect of headbox adjustments in real-time is investigated with respect to the above-mentioned parameters. It was found that, for the specific camera model and tracking algorithm, one or two headbox adjustments per second, as would normally be the case during recording of human participants, could be tolerated in favour of a higher framerate. The effect of adjustment of the recording window can be reduced by using a larger recording window at the cost of the framerate.

Using Smooth Pursuit Calibration for Difficult-to-Calibrate Participants.

Although the 45-dots calibration routine of a previous study ( 2) provided very good accuracy, it requires intense mental effort and the routine proved to be unsuccessful for young children who struggle to maintain concentration. The calibration procedures that are normally used for difficult-to-calibrate participants, such as autistic children and infants, do not suffice since they are not accurate enough and the reliability of research results might be jeopardised. Smooth pursuit has been used before for calibration and is applied in this paper as an alternative routine for participants who are difficult to calibrate with conventional routines. Gaze data is captured at regular intervals and many calibration targets are generated while the eyes are following a moving target. The procedure could take anything between 30 s and 60 s to complete, but since an interesting target and/or a conscious task may be used, participants are assisted to maintain concentration. It was proven that the accuracy that can be attained through calibration with a moving target along an even horizontal path is not significantly worse than the accura-cy that can be attained with a standard method of watching dots appearing in random order. The routine was applied successfully for a group of children with ADD, ADHD and learning abilities. This result is important as it provides for easier calibration - especially in the case of participants who struggle to keep their gaze focused and stable on a stationary target for long enough.

Fixation identification: the optimum threshold for a dispersion algorithm.

It is hypothesized that the number, position, size, and duration of fixations are functions of the metric used for dispersion in a dispersion-based fixation detection algorithm, as well as of the threshold value. The sensitivity of the I-DT algorithm for the various independent variables was determined through the analysis of gaze data from chess players during a memory recall experiment. A procedure was followed in which scan paths were generated at distinct intervals in a range of threshold values for each of five different metrics of dispersion. The percentage of points of regard (PORs) used, the number of fixations returned, the spatial dispersion of PORs within fixations, and the difference between the scan paths were used as indicators to determine an optimum threshold value. It was found that a fixation radius of 1 degrees provides a threshold that will ensure replicable results in terms of the number and position of fixations while utilizing about 90% of the gaze data captured.