Incorporation of robotic automated transcranial Doppler to screen for patent foramen ovale (PFO) and quantify right-to-left shunt severity in the evaluation of ischemic stroke patients for etiology and PFO management.

Background: Right-to-left shunt (RLS) associated with patent foramen ovale (PFO) is common among cryptogenic strokes. Current diagnostic tools have limitations. Transthoratic echocardiography (TTE) is not as sensitive as Transesophageal echocardiography (TEE), TEE is invasive, and manual transcranial Doppler (TCD) requires trained staff to operate. Robotic automated TCD (raTCD) may be feasible and comparable to manual TCD. The study's purpose was to determine the rate of RLS detection using raTCD and combine the Spencer Logarithmic Scale (SLS) with the Risk of Paradoxical Embolism (RoPE) to identify patients at risk of PFO associated stroke or TIA.

Methods: This single-center retrospective cohort study included adult patients (≥18 y/o) admitted from December 2021 to December 2022 with a stroke or transient ischemic attack. Those with no bone window or stroke mimics were excluded. Patients with an RLS on raTCD received a second scan at the physician's discretion. The SLS combined with the RoPE score was used to generate a modified screening PFO-Associated Stroke Causal Likelihood (msPASCAL) classification.

Results: Of 212 patients who received raTCD, the mean age was 56, 14% were >65 years old, most were white (72%), predominantly male (59%), 52% had cryptogenic strokes, and 59% had an RLS. Most patients were able to perform Valsalva (89%) during raTCD. Of those with an RLS, 56% had an SLS of 1-2, while 44% had an SLS of 3-5. There were no significant differences in characteristics by SLS. Most patients with SLS grades 1-2 were classified using msPASCAL as unlikely to have PFO as stroke etiology ( = 55, 44%). A small number of large SLS grades 3-5 were considered probable for having a PFO-associated stroke while the rest were classified as possible ( = 38, 30.4%). Eight patients with positive RLS on raTCD had a negative TTE with bubbles; most of those had small RLS on raTCD ( = 5, 63%) or could not Valsalva due to sedation (( = 6, 75%).

Discussion: This study supports the feasibility of utilizing raTCD for RLS detection. The modified screening PASCAL classification can be generated for RLS patients and may be used to guide subsequent evaluation and management.