An echo from the past: open access repository of over 10,000 annotated Doppler audio recordings of venous gas emboli.

A total of 10,099 Doppler ultrasound recordings were compiled into an open-access repository released under a public domain license.

• Audio tapes with recorded Doppler data were converted to digital files, then cut into individual recordings and matched with their metadata
• The dataset includes subject and pressure profile information
• The database has been released under a public domain license for further use
• Recordings originated from Doppler ultrasound measurements recorded since the 1970s across the world

The dataset included recordings from up to 311 divers, all male, with a median age of 31.5 years among the 170 identified divers.

• Total divers: n = ≤311, with 170 identified and ≤141 unidentified
• All divers were male
• Median age was 31.5 years among the 170 identified divers
• Breathing gases included air, nitrox, and heliox

The dives covered a range of maximum depths from 24 m (80 feet) to 91.4 m (300 feet).

• Minimum maximum depth was 24 m (80 feet)
• Maximum depth reached was 91.4 m (300 feet)
• Breathing gases included air, nitrox, and heliox

Doppler measurements were taken between 2 minutes and 594 minutes post-dive, with a median time of 52 minutes.

• Earliest measurement was 2 minutes post-dive
• Latest measurement was 594 minutes post-dive
• Median measurement time was 52 minutes post-dive

Decompression sickness (DCS) was noted in only 12 individuals in the dataset.

• DCS occurred in 12 individuals out of ≤311 divers
• The dataset centred around lower VGE loads (Spencer Grades 0, I, and II)
• Spencer grading was used to classify venous gas emboli load

The large dataset enabled development of a deep learning algorithm capable of grading bubble loads without a human operator.

• Audio signals and their Doppler grades were processed further for suitability to train an algorithm to identify VGE
• The algorithm can grade bubble loads without a human operator
• The dataset represents a potential resource for training deep learning algorithms to recognise and grade venous gas emboli