Abstract
Background: AI is being used increasingly in image interpretation tasks. There are challenges for its optimal use in reporting environments. Human reliance on technology and bias can cause decision errors. Trust issues exist amongst
radiologists and radiographers in both over-reliance (automation bias) and reluctance in AI use for decision support. A checklist, used with the AI to mitigate against such biases, may optimise the use of AI technologies and promote good
decision hygiene.
Method: A checklist, to be used in image interpretation with AI assistance, was developed. Participants interpreted 20 examinations with AI assistance and then re- interpreted the 20 examinations with AI and a checklist. The MSK images
were presented to radiographers as patient examinations to replicate the image interpretation task in clinical practice. Image diagnosis and confidence levels on the diagnosis provided were collected following each interpretation. The
participant perception of the use of the checklist was investigated via a questionnaire.
Results: Data collection and analysis are underway and will be completed at the European Congress of Radiology in Vienna, March 2023. The impact of the use of a checklist in image interpretation with AI will be evaluated. Changes in
accuracy and confidence will be investigated and results will be presented. Participant feedback will be analysed to determine perceptions and impact of the checklist also.
Conclusion: A novel checklist has been developed to aid the interpretation of images when using AI. The checklist has been tested for its use in assisting radiographers in MSK image interpretation when using AI.
radiologists and radiographers in both over-reliance (automation bias) and reluctance in AI use for decision support. A checklist, used with the AI to mitigate against such biases, may optimise the use of AI technologies and promote good
decision hygiene.
Method: A checklist, to be used in image interpretation with AI assistance, was developed. Participants interpreted 20 examinations with AI assistance and then re- interpreted the 20 examinations with AI and a checklist. The MSK images
were presented to radiographers as patient examinations to replicate the image interpretation task in clinical practice. Image diagnosis and confidence levels on the diagnosis provided were collected following each interpretation. The
participant perception of the use of the checklist was investigated via a questionnaire.
Results: Data collection and analysis are underway and will be completed at the European Congress of Radiology in Vienna, March 2023. The impact of the use of a checklist in image interpretation with AI will be evaluated. Changes in
accuracy and confidence will be investigated and results will be presented. Participant feedback will be analysed to determine perceptions and impact of the checklist also.
Conclusion: A novel checklist has been developed to aid the interpretation of images when using AI. The checklist has been tested for its use in assisting radiographers in MSK image interpretation when using AI.
| Original language | English |
|---|---|
| Pages | 85, P103 |
| Publication status | Published (in print/issue) - 23 Jun 2023 |
| Event | UK Imaging & Oncology Congress 2023 - Liverpool, United Kingdom Duration: 5 Jun 2023 → 7 Jun 2023 |
Conference
| Conference | UK Imaging & Oncology Congress 2023 |
|---|---|
| Abbreviated title | UKIO |
| Country/Territory | United Kingdom |
| City | Liverpool |
| Period | 5/06/23 → 7/06/23 |
