Validation Of The Automated Electronic Microemboli Detection System

Letter to the authors:

I want to congratulate the authors for their study of the “Emboli Detection System” (EDS) during carotid surgery [1]. I appreciate their efforts to prevent strokes in patients scheduled for carotid surgery. As the designer of the EDS, I would like to comment on their final conclusion. The EDS was designed as an assistive tool for human experts for rapidly identifying the presence of micro-embolic signals (MES) in TCD time series. The EDS was never intended as a tool to replace human experts. Since its introduction in 2008, rapid and reliable identification of MES has been performed in hundreds of patients with the help of the EDS in numerous European Hospitals. Especially in clinical settings with a low frequency of cerebral embolism, such as seen in patients with asymptomatic carotid artery stenosis, neurologists, clinical neurophysiologists and vascular technologists value the assistance of the EDS. 

The EDS helps the human expert, including the authors, by offering different features, such as an intensity threshold function, audio-replay function and detailed zoom function of the TCD signal. Moreover, the EDS includes a classifier with a very high total accuracy (> 96 %), allowing the human expert rapid identification of those events that might be true MES [2]. However, this classifier is unfit for use as a stand-alone classifier, as reported by the authors. For this reason, every installed EDS has a disclosure statement in its Instructions For Use (IFU). The EDS instructions state that the EDS can never be used as an independent system to classify MES. In fact, although expressed differently, the authors scientifically confirmed the disclosure statement of the EDS.

Unfortunately, the authors did not mention in their article that the EDS was developed as an assistive tool for human experts. As a result, the article gives the reader the impression that the EDS is not working according to specifications. This is patently not the case. In fact, it is incorrect to conclude that the EDS has failed to fulfill the criteria of automatic embolus detection without pointing out that the EDS was not designed for this purpose. The reader should have been informed of this matter.

• References

• 1 Leunissen T. et al. Ultraschall in Med 2017; DOI: 10.1055/s-0043–106737. [Epub ahead of print]
• 2 Keunen R. et al. J Med Eng Technol 2008; 4: 296-304 . doi:10.1080/03091900701541265