University of Twente Student Theses
Transducer selection & protocol compliance detection algorithm design as a basis for automated point-of-care lung ultrasound screening of tuberculosis.
Room, J.T. (2023) Transducer selection & protocol compliance detection algorithm design as a basis for automated point-of-care lung ultrasound screening of tuberculosis.
|
PDF
17MB |
| Abstract: | Tuberculosis(TB) is one of the most prevalent diseases in the world. Most TB infections occur in third world countries, where the diagnosis of TB is difficult due to limited access to healthcare, funds, and infrastructure. To make early diagnosis possible in these areas, a cost-effective point-of-care device screening tool in combination with a follow-up test would be ideal. A pointof-care ultrasound (POCUS) device for automated TB screening could increase the number of TB cases detected in areas of interest. These cases can then be sent to the hospital to confirm whether the symptoms are TB. For the device, a standardized easy-to-learn scan protocol with an affordable point-of-care ultrasound transducer can be combined with computer-aided detection to eventually exclude the health worker from the screening process. This thesis takes the first steps toward the design of a cost-effective standardized computerassisted lung ultrasound (LUS) screening tool for TB. By evaluating different commercial affordable POCUS transducers on their usability and their ability to detect TB-related LUS signs via an assessment matrix, B-line phantom, and quality assurance test. The highest scoring transducer was used for a clinical study on TB detection with LUS using a designed standard scan procedure. Usable scan lines for diagnosis in LUS are operator dependent. The quality of the recording is influenced by the scan angle toward the pleural line and the proper contact between the skin and the transducer. Therefore, in the second part of this thesis, clinical study data was used to design a shadow peak-based protocol compliance algorithm that is capable of checking whether a scan line is performed correctly. This algorithm was designed to guide an inexperienced user during the scan procedure to ensure proper data collection. The data collected using this protocol compliance protocol can be used to train a deep learning algorithm for computer-aided screening of TB. The ideal transducer out of the tested transducers for a cost-effective POCUS screening tool for the detection of TB with LUS is the Clarius C3 HD3 multipurpose transducer. The protocol compliance algorithm designed is capable of accurately selecting sufficiently recorded videos of scan lines with a sensitivity of 100% out of a data set containing 30 videos of scan lines. The algorithm made a correct judgment for 87% of the input scan lines with the greatest error in the selection of insufficient scans resulting in a specificity of 60%. The algorithm shows promise, and the physical principle seems to be usable for the selection of sufficient scan lines. However, the algorithm should be re-evaluated using a larger, more representative data set for training and testing to achieve a better judgment of its performance. The selected transducer and the protocol compliance algorithm design are the first steps toward a cost-effective standardized computer-assisted LUS screening tool for TB. |
| Item Type: | Essay (Master) |
| Clients: | Delft Imaging, Den Bosch, Netherlands |
| Faculty: | TNW: Science and Technology |
| Subject: | 33 physics, 42 biology, 44 medicine, 50 technical science in general, 54 computer science |
| Programme: | Biomedical Engineering MSc (66226) |
| Link to this item: | https://purl.utwente.nl/essays/95386 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page