Artificial Intelligence Aided Surgical Planning System for Malignant Bone Tumour Excision with Intraoperative Safety Monitoring

Professor Zhang, Teng Grace   (Principal Investigator (PI))

24

2024-03-01

2298730

Artificial Intelligence Aided Surgical Planning System for Malignant Bone Tumour Excision with Intraoperative Safety Monitoring

Artificial Intelligence, Malignant, Bone Tumour, Excision, Intraoperative Safety

Orthopaedics/Traumatology

ITS/286/22FP

Innovation and Technology Support Programme

2024

On-going

The bone tumour excision is complicated and risky that requires precise preoperativeplanning and intraoperative navigation. However, manual surgical planning highly relieson specialists’ experience and subjective judgment, which is inefficient and inconsistent.Besides, the existing surgical navigation focuses on trajectory guidance but lacks safetymonitoring.Through this project, we aim to develop and validate the first surgical auto-planning andsafety monitoring system for bone tumour excision. Our objectives include: 1) establish adataset containing multi-modality images with specialists’ annotation; 2) develop a multitissue3D reconstruction pipeline via semantic segmentation and multi-modalityregistration; 3) develop a surgical auto-planning and safety monitoring system; 4) conducttechnical and mock clinical validation on the developed system.We have published three ground-breaking methodologies, 1) the Spine-GFlow for robustand unsupervised multi-tissue segmentation; 2) the SpineGEM for model training using asmall dataset; 3) the SpineHRNet key-point detection pipeline for multi-modalityregistration, at CMIG, MICCAI, and Lancet: EClinicalMedicine respectively.The deliverables of this project enable fast and accurate surgical planning as well as safenavigation that improves the general wellness of all patients suffering from bone tumours.They and their families will benefit from fast recovery and improved prognosis.