With the expansion of modern communication technologies, many applications have been proposed in medicine, in order to improve the access of care in situations where direct interaction between patient and health care provider is impossible, or strongly restricted [1]. If the initial development of these applications was largely supported by military administrations, common civil interest is actually well recognized [2], [3], [4].
Current applications of telemedicine are distributed in all the fields of medicine, from nursing [5] to emergency work [6], from dermatology [7] to medical imaging [8], [9]. Nevertheless the common clinical use of telemedicine largely varies from one medical specialty to one other, with different degrees of acceptance from the users [10].
Teleradiology is one of the most mature applications of telemedicine. Its first clinical application, named "telognosis" (for "teleoroentgen diagnosis"), was described in 1950 [11]. Indeed, related to the conventional radiological workflow, teleradiology does not significantly modify the current practice modality. For a large part of their activity, imaging specialists are heavily focused on images rather than on face-to-face interaction with the patient. For many imaging technologies the diagnostic accuracy doesn't significantly change if the examination takes place next door or kilometers away [12], [13].
On account of medical specialists' shortage, teleradiology becomes more and more frequently used in small or medium size hospitals for image interpretation, for routine daily work or for emergency [14], [15], [16]. The 1999 survey from the American College of Radiology demonstrated that the most common use of teleradiology (76-99 % of radiology practices with teleradiology systems) was for preliminary on-call findings in emergency settings [17]. If remote image interpretation doesn't give rise to problems with plain films, computed tomography (CT) or magnetic resonance imaging (MRI) exams [18], no satisfactory and universally recognized solution exists at this time for ultrasound (US) [19]. This is certainly related to one fundamental difference of sonography in comparison with other imaging modalities, the same one that makes sonography particularly operator dependant: the need of direct interaction of an experienced sonologist with the patient to obtain clinically useful information.
Different technical solutions to the dilemma of how to generate meaningful ultrasound images when an experienced sonologist is away have been proposed. These solutions either need specific competences on site or very specific expensive material, or both. They may be subdivided into three groups [20].
The first group is closely related to the paradigm of direct and active involvement of the physician in real time [21]. In order to assure the remote control of the probe positioned on the patient, some authors have advocated the use of a robot [22], [23]. With the robot, the remote expert performs the examination more or less as if he/she was himself by the side of the patient. This concept needs dedicated materials for remote control and videoconference facilities, in order to communicate with the local operator driving the robot at the patient site. The local operator has to be instructed for robot manipulations, but not for sonography.
A second group use communication facilities to send conventional images acquired on site by people with brief sonography training to a skilled physician located at the remote site. These may be stored images with asynchronous interpretation [24], [25], [26] or real-time interactive video systems for telementoring of less experienced colleagues on site through videoconference systems [27], [28], [29], [30], [31]. This design may be used with commonly available ultrasound systems and simple communication facilities for store and forward systems, but requires broad-bandwidth network for simultaneous transmission of voice and video of the participants, ultrasound images of the patient (still images or real-time video), written information among the participants and data for the coordination between the different computers if telementoring is taken up.
The third group promotes the acquisition of tridimensional ultrasound data set by relatively unskilled operators on site with remote interpretation of volume data by an experienced sonologist. This interpretation may be performed off-line [32], [33] or online with dedicated soft and hardware [34], [35].
Synchronous online interpretation needs to have an expert available during the performance of remote examination in order to interpret the ultrasound images or films and guide the operator proceeding to the examination on site. This model may only be successful in sufficiently large structures, which may have a sufficient number of sonologists, in order to have continuously available experts for remote examination supervision [16], [36].
Asynchronous off-line interpretation means that a time delay appears between the images acquisition, the transmission to the expert and his interpretation and answer to the remote location [37]. This kind of process is not appropriate for absolute emergency examinations, such as focused abdominal sonography for trauma (FAST) or if the local operator is unable to independently provide interpretable image data sets.
Nowadays even with a very short delay between acquisition and remote interpretation, asynchronous reading represents actually the larger way of using teleradiology, particularly for CT and MRI interpretation [38]. Asynchronous processes permits the insertion of remote workload in the local workflow with minimal interferences. In order to use telesonography in the same way, it is absolutely necessary to release it from the local operator dependency. Remote control of a robot is one solution, but needs the local collaboration of a technologist trained for robot manipulations. Volume ultrasound is another solution that only needs the help of a local co-worker who put the volumetric probe on predetermined points on the patient, in order to record the volumetric data sets. No particular training is required in that case.
Radiologists are getting used to manipulate and interpret large amounts of images from the multidetector CT (MDCT) or MRI [39], [40]. Retrieved US images from volume data are equivalent to these large images sets, but dedicated workstations are actually lacking. When available, reconstruction of diagnostic US images from tridimensional acquisition will be accessible to adequately trained sonologists. Further developments are needed to obtain these user friendly efficient sonographic workstations.
In conclusion, as volume US acquisition seems to be rather independent to the skill of the local operator, asynchronous remote interpretation of reconstructed images from volume data sets appears to be a well appropriate technical solution for telesonography, providing that suitable workstations for post-processing will be available.
Prof Jean-Yves Meuwly
Department of Diagnostic and Interventional Radiology
University Hospital Lausanne
1011 Lausanne
Switzerland
