Potential of Simulators in Ultrasound Diagnostics
Artificial intelligence (AI) and ultrasound simulators are being used increasingly
in medical imaging. With the rapid advent of AI and phantoms in the field of medicine,
the question arises as to what influence they have on ultrasound diagnostics. If you
ask ChatGPT 3.5 about the efficacy of ultrasound simulators, it will tell you, among
other things, that ultrasound simulators are invaluable in medical education and in
the training of physicians and medical staff, and that aspiring physicians can learn
the basics of ultrasound imaging without jeopardizing the health and well-being of
real patients and without the need for expensive equipment. Even though this seems
somewhat exaggerated at the moment, the advantages of ultrasound phantoms and automated
image analysis are evident. One focus of the so-called “SonoTrainer” is its use in
the training and continuing education of physicians. The simulators available today
already cover various interdisciplinary fields of application. This includes both
diagnostic imaging and interventional procedures for training purposes as well as
for diagnosis, setting measurement points, device calibration, and quality control.
So-called “tissue-mimicking phantoms” mimic the acoustic properties of human tissue
to simulate ultrasound images. “Needle insertion phantoms” support learning of ultrasound-guided
procedures, such as amniocentesis or tumor biopsy. “Breast phantoms” simulate breast
tissue imaging and are used for training and continuing education in breast sonography.
“Obstetric gynecological ultrasound phantoms” are also increasingly being discussed
as an education and training tool in prenatal diagnostics and vaginal sonography.
In view of the fact that continuing education is facing major challenges in many areas
of medicine due to scarce institutional, personnel and financial resources, the integration
of ultrasound simulators into the medical curriculum can be used as a complementary
part of a structured learning program during medical studies, or for the continuing
medical education of physicians. Examples include ultrasound training as part of medical
studies. In this context, a study from Lausanne comprising three hours of theoretical
and practical content, incl. FAST/eFAST examinations and looking for free fluid on
an ultrasound simulator, showed that 89 % of students were in favor of using an ultrasound
simulator. It was also shown, however, that 53 % of the students already felt they
were competent to perform an ultrasound examination of the abdomen after just three
hours of training. This prompted me to reflect on my role as a teacher, and highlighted
the importance of balanced planning of course content and the critical use of newer
methods [1]. These experiences have shown how important it is, notwithstanding the excitement
about this technology, to also convey the complexity of sonography and the limitations
of using a phantom.
In prenatal diagnostics as well, ultrasound simulators and AI support learning and
recognition of normal sonoanatomy and standardized biometry, as well as visualization
of fetal malformations. A meta-analysis conducted by the German Institute for Quality
and Efficiency in Health Care (IQWiG) showed a positive association between the qualifications
and experience of the examiner and the detection rate of fetal anomalies [2].
Recent studies show that structured learning using a phantom can raise the standard
of ultrasound examination in prenatal diagnostics and improve measurement accuracy.
Training on the simulator also enables faster learning and visualization of standard
fetal structures, including fetal echocardiography and fetal anomalies [3]
[4]
[5]. In a multicenter study published in this edition, Zhao et al. show that the simulator-based
obstetric ultrasound competency assessment tool (OUCAT) has good reliability and validity
in assessing ultrasound skills in obstetrics, and can be used to assess the competence
of ultrasound examiners. They were also able to demonstrate that the competence of
experts was significantly better than that of experienced trainees, and experienced
trainees were significantly better than beginners. In this study, the OUCAT comprised
123 elements, 117 of which were able to clearly distinguish between beginners and
experts (p < 0.05).
The INVUS phantom, also presented in this edition by Seitzinger et al., enables standardized
and realistic training in ultrasound-guided procedures. The study participants included
inexperienced (n = 40) and experienced ultrasound examiners (n = 41). Of a total of
81 ultrasound examiners, 73 participants rated the visualization of the lesions as
a realistic representation and 86 % (70/81) considered the phantom to be of high clinical
significance for learning ultrasound-guided puncture procedures [7].
With regard to the future of the ultrasound machines themselves, it is expected that
the resolution will increase and the image quality will improve. AI algorithms will
increasingly lead to the automation of image analysis, and perhaps also to providing
real-time support in imaging. It should be noted that the data collected so far, while
promising, is retrospective and cannot be applied directly to clinical work. In addition
to larger training datasets, there is a need for validation studies, which may be
particularly helpful for inexperienced ultrasound examiners [8].
This is of relevance since, for example, the number of diagnostic puncture procedures,
such as chorionic villus sampling or amniocentesis in prenatal diagnostics, has rapidly
decreased due to the optimization of non-invasive screening for trisomies [9]. As a consequence of the declining puncture rates in prenatal diagnostics, not enough
physicians are handling the number of cases required to become qualified in this technique
[9]. The phantoms for biopsy training could potentially facilitate learning as an additional
module; however, this should first be scientifically evaluated.
Combined with rapid developments in AI, the newest generation simulators will not
replace core medical skills – but they will help to better structure training and
continuing education, and better prepare physicians for clinical applications in their
work. For example, a first step would be to determine what proportion of ultrasound
simulator training should be included in different courses. Some of the required number
of ultrasound examinations could be performed on the phantom. It would also be useful
to set up workstations with ultrasound simulators in the clinics that can be used
by trainees and for ongoing education. The simulators already enable users to practice
various scenarios and obtain feedback. The SonoTrainer thus paves the way for new
perspectives in mentoring and for certification programs. The importance of ultrasound
simulators in the future will depend on how realistic the simulator is. Realistic
tissue models and imaging modules will support the achievement of learning objectives.
At the same time, it is necessary to perform a scientific evaluation of the efficacy
of this equipment, which can be expensive to purchase, in terms of meeting the clinical
requirements. And considering that medical tasks are not limited to diagnostic imaging
and clinical evaluation of the images, communication skills will also continue to
be a crucial part of the physician’s role – accordingly, simulators and AI will not
replace medical tasks, but will challenge and promote them.
