Proposed Topic (Most preferred): :
Research and Innovations (new projects / technology / innovations / service models)
Authors (including presenting author) :
Lam TP(1), Ng TM(1), Chan PK(1), Kan YL(1)
Affiliation :
(1)Department of Radiology, Hong Kong Children's Hospital
Introduction :
Bone age is an essential index for assessing the growth trend of a child. Traditionally the assessment is performed with manual comparison of the left hand radiograph with an image atlas by radiologists. This is a time consuming process and inter-rater variability is inevitable. The project team had investigated the feasibility of deploying AI tool determine the bone age from digital radiographs. An automised bone age assessment solution was identified to be suitable for the purpose. After deliberation, pilot run and validation, the tool was implemented in Department of Radiology, Hong Kong Children’s Hospital in 2021. The project was further extended to Radiology of hospitals in KCC and then other clusters. The implementation to Radiology of all HA hospitals was completed in 2023.
Objectives :
-Identify a potential product in the market.
-Perform pilot run and validation test to prove the security and accuracy of the target products
-Implement into clinical practice
Methodology :
The team gone through a process of selecting a target product, understand the technical requirements as well as possible risk. With clarifications with potential suppliers, support from higher management is sough for balancing clinical benefit, technical feasibility, information security, financial sustainability and also future scope of the project.
With reference to experience from foreign hospitals network, one commercial solution is identified. Technical feasibility of connecting to HA network and necessary precautions dealing with patient data privacy and network security was evaluated with IT team and a solution was proposed. The proposed solution was put forward to seek support from higher management for a pilot run and resource approval. During the pilot run, a validation test was performed for verifying whether the AI analysis aligns with the assessment performed by radiologists. The project was supported to put into clinical use after the evaluation of the pilot study. The first implementation site is HKCH.
To further rollout, liaison with stakeholders in all hospitals become the essential task. Clinicians, radiologists, PACS Administrators, radiographers and IT department of all clusters were needed be contacted. The project was implemented in 4 phases and finally all HA hospitals joined in 2023.
Result & Outcome :
Workflow in Radiology was changed while all bone age images can now be read by AI tool automatically. The analysis result is uploaded to ePR for all clinicians’ reference.
