Abstract
When considering the various medical conditions NASA is preparing to encounter in along duration space mission to Mars, they must assess the value of sending various personnel and decide which aspects of training could be most helpful . When specifically considering ultrasound diagnostic procedures, there is general understanding that ultrasound would be useful, but there has been limited analysis performed on the risk reduction of sending ultrasound diagnostics to space on long duration space missions. This paper discusses the specific utility of automated ultrasound diagnostics, for conditions predefined by NASA as likely to occur. Both collecting and analyzing the images currently require clinical training and expertise which could be offset by an automated system. As medicine has become increasingly specialized, a tool with generalist expertise could aid astronauts in a timely fashion to collect and diagnose conditions critical to preserving resources and survival in space.
Autonomous ultrasound image collection and detection was explored separately for these enumerated conditions by NASA. Recommendations were based on timing of onset and duration of conditions, efficacy and utility of diagnosis for ultrasound, and training capability that could be offloaded to an autonomous system versus crew expertise requirements for launch. Of the list of conditions from NASA, 20 were identified as diagnosable, in part, from ultrasound, either the condition itself or for ruling out another condition with similar symptoms. The efficacy of this diagnosis for clinicians in recent literature was explored for each condition as autonomous systems are generally trained with data from clinician identified images, making autonomous systems only, at best, as good as the clinician data used to train them. This, along with current research in a few particular conditions, generated the data to help select the conditions diagnosable by ultrasound to direct support for technological innovation in autonomous detection. Additionally, autonomous ultrasound collection is currently not available, outside of limited areas of the body with significant clinician oversight. Supporting applied research in this field could additionally offload medical training and standardize images collected, a current shortcoming of ultrasound scans today.