uSINE-PAMS Artificial Intelligence Driven, Ultrasound-guided Lumbar Puncture to Improve Procedural Accuracy—A Pilot Usability Study
Xuling Lin1, Ding Fang Chuang1, Liqing Fu1, Seyed Ehsan Saffari2, Christen Lim1, Yu-Lin Wong3, Joanne Yuen1, Valerie Teh1, Aynul Marliya1, Carissa Lam1, Christopher Seet1
1National Neuroscience Institute, 2Duke-National University of Singapore, 3Tan Tock Seng Hospital
Objective:
To test the usability of uSINE-PAMS for Lumbar Puncture (LP)
Background:
Lumbar puncture (LP) is an invasive diagnostic procedure. The traditional landmark-based method is associated with a failure rate of up to 50%, leading to complications, diagnostic delay and increased healthcare costs. Ultrasound-guided LP has improved the procedural success rate. However, ultrasound use has not been widespread due to inadequate neuraxial-ultrasound training and inability to reliably translate information derived from ultrasonography to LP procedure. uSINE-PAMS was designed to address these challenges: uSINE is a machine-learning software to aid in ultrasound-guided LP; PAMS is a two-part hardware to translate data from ultrasonography to accurate needle insertion.
Design/Methods:
A single-arm, prospective study of uSINE-PAMS was planned. Two trained neurologists performed the uSINE-PAMS-guided LP. Patients slated to undergo LP for their clinical care and able to provide informed consent were recruited. Number of successful first-pass, number of needle redirections, peri-procedural pain score and complications, cerebrospinal fluid (CSF) red blood cells (RBC) and patients’ demographics were collected. Descriptive statistics was performed.
Results:
Ten patients (6 men, 4 women) were recruited from May to August 2023. Median age was 43 (range 30-63) years; median body mass index was 24.5 (16.1-34.3) kg/m2. uSINE-PAMS-guided LP was successfully performed with first pass in 8 patients, 2 of whom required needle redirections. The remaining 2 patients had LP performed successfully at second and third attempts respectively; first pass failure in these 2 patients was felt to be related to initial inexperience with PAMS use. Median pain score (range 0-10) was 4 and there were no LP-related complications. Three patients had CSF RBC >5 (range 9-4200) cells/uL.
Conclusions:
This first-in-human study of PAMS showed that uSINE-PAMS is feasible for LP. First-pass success rate was promising at 80% in this pilot study. A phase 2 study of uSINE-PAMS-guided LP is ongoing. (NCT05824546)