Browsing by Author "Douglas, Tania Samantha"

Now showing 1 - 2 of 2
  • Thumbnail Image
    Item
    Open Access
    Exploring the visualisation of the cervicothoracic junction in lateral spine radiography using high dynamic range techniques
    (2019) Kung, Quik; Douglas, Tania Samantha; Dawson, Sarah
    The C7/T1 junction is an important landmark for spinal injuries. It is traditionally difficult to visualise in a lateral X-ray image due to the rapid change in the bodys anatomy at the level of the junction, where the shoulders cause a large increase in attenuation. To explore methods of enhancing the appearance of this important area, lateral radiographs of a shoulder girdle phantom were subjected to high dynamic range (HDR) processing and tone mapping. A shoulder girdle phantom was constructed using Perspex, shoulder girdle and vertebral bones and water to reproduce the attenuation caused by soft tissue. The design allowed for the removal of the shoulder girdle in order for the cervical vertebrae to be imaged separately. HDR was explored for single and dual-energy X-ray images of the phantom. In the case of single-image HDR, the HDR image of the phantom without water was constructed by combining images created with varying contrast windows throughout the contrast range of an X-ray image. It was found that an overlap of larger contrast windows with a lower number of images performed better than smaller contrast windows and more images when creating an HDR to be tone mapped. Poor results on the phantom without water precluded further testing of single-image HDR on images of the phantom with water, which would have higher attenuation. Dual energy HDR image construction explored images of the phantom both with and without water. A set of images acquired at lower attenuation (phantom without water) was used to evaluate the performance of the various tone mapping algorithms. The tone mapping was then performed on the phantom images containing water. These results showed how each tone mapping algorithm differs and the effects of global vs. local processing. The results revealed that the built-in MatLab algorithm, based on an improved Ward histogram adjustment approach, produces the most desirable result. None of the HDR tone mapped images produced were diagnostically useful. Signal to noise ratio (SNR) analysis was performed on the cervical region of the HDR tone mapped image. It used the scan of the phantom without the shoulder girdle obstruction (imaged under the same conditions) as a reference image. The SNR results quantitatively show that the selection of exposure values affects the visualisation of the tone mapped image. The highest SNR was produced for the 100 - 120 kV dual energy X-ray image pair. The study was limited by the range of HDR image construction techniques employed and the tone mapping algorithms explored. Future studies could explore other HDR image construction techniques and the combination of global and local tone mapping algorithms. Furthermore, the phantom can be replaced by a cadaver for algorithm testing under more realistic conditions.
  • Thumbnail Image
    Item
    Open Access
    User-interface design and evaluation in a mobile application for detecting latent tuberculosis
    (2019) Farao, Jaydon Ethan; Douglas, Tania Samantha; Malila, Bessie; Mutsvangwa,Tinashe
    Treatment and monitoring of tuberculosis have been met with various interventions to reduce its prevalence. One such intervention, to detect and prevent latent tuberculosis infection (LTBI), is the tuberculin skin test (TST), for which an induration response on a patient’s arm is an indication of LTBI. The test requires the patient to return to a clinic 48 to 72 hours after TST administration for assessment of the response. This is a challenge because of financial and accessibility obstacles, especially in under-resourced regions. A mobile health (mHealth) application (app) has been developed for remote assessment of the response to the TST. The previous version of the LTBI screening app, however, had usability limitations. The app is intended for use by patients and healthcare workers; thus, ease of use is important. There is a lack of literature on the usability of mHealth apps, especially in under-resourced settings. In this project, the user interface of the app was redesigned and tested. The Information Systems Research (ISR) framework was integrated with design thinking for this purpose. The project included creating mock-ups of the interface which were iteratively prototyped with ten student participants, adjusted, and assessed according to the user feedback. Thereafter, the Android Studio software was used to adjust the user interface based on the insights gained through the progression of prototypes. The usability of the updated app was tested and assessed with ten healthcare workers at a community health clinic in Khayelitsha in Cape Town, South Africa. Data collection and analysis comprised both qualitative and quantitative methods. Observations, the “think aloud” approach, and the post-study system usability questionnaire were used for data collection. Student participants highlighted various usability limitations of the app during each iteration. The major usability limitations included: the complex image capture protocol, misunderstanding of instructions, and time taken to capture images. Engagement with students allowed for improvement of the app interface and enabled adequate preparation for testing in the field with end-users. Furthermore, improving the app interface before engaging with healthcare workers, enabled context specific limitations that would affect the usability of the app, to be explored during the field testing. These included safety concerns when using the app and the privacy of health information. Future work should explore how these concerns, as well as other social factors, affect usability. Furthermore, improving the image capture protocol is required for improving the usability of the app.