Robert Lorentsson: New method for detecting ultrasound equipment errors
Many ultrasound examinations are conducted using equipment with malfunctions, often without anyone’s knowledge.
In his doctoral thesis, Robert Lorentsson reveals that these errors can impact diagnosis, especially when the defects are substantial. He has developed a novel method to detect faulty ultrasound probes through automated analysis of clinical images.
ROBERT LORENTSSON
Dissertation Defense: September 15, 2023
Doctoral Thesis: How does the technical status of medical ultrasound equipment affect image quality? Studies based on human observer experiments and a novel method for automatic detection of defective transducers
Research Area: Medical Radiation Science
Sahlgrenska Academy, The Institute of Clinical Sciences
What is the focus of your research?
”My research aims to evaluate how the technical condition of an ultrasound machine affects the clinical image quality. In an observational study, four experienced radiologists assessed ultrasound images produced both with flawless equipment and with defective probes,” says Robert Lorentsson, a biomedical engineer working in the field of medical physics and biomedical engineering at Sahlgrenska University Hospital.
What led you to study faulty probes?
”In fact, annual equipment checks reveal defects in up to 27 percent of the probes. Additionally, the images used in the study involving faulty probes were obtained from clinical use. For 75 percent of the defective probes, the overall image quality was assessed as poorer than that of the faultless ones.”
Revealing defects through dark regions
What are the consequences when ultrasound examinations are conducted with suboptimal equipment?
”It could significantly impact the diagnosis. In the observational study with faulty probes, one question was how confident the observer was that the image artifact resulting from the defect could affect the diagnosis. In 19 percent of the assessments of images from the defective probes, the answer was that the observer was ‘confident that the artifact could impact the diagnosis’.”
How can the usage of defective ultrasound equipment be minimized?
”In my doctoral project, we developed a new method for early detection of faulty probes. This method involves analyzing clinical ultrasound images stored for documentation. When the most recent images are analyzed, defects can be detected. The method identifies darker regions in the images, a characteristic of those produced with defective probes.”
Automated analysis by computer
How long does it take to detect malfunctions through image analysis?
”There is a delay of 75-150 images before the defects are identified, which translates to up to two weeks of usage for a frequently used probe. The method can handle both linear and curved probes. Everything is managed by a computer that automatically imports the images and analyzes them for multiple probes simultaneously. A significant advantage of this method is that it doesn’t require access to the ultrasound equipment, thus not disrupting operations.”
”Key insight that diagnoses can be affected”
What are the most crucial findings in your thesis and what practical value do they offer?
”The realization that defective equipment can even impact diagnoses in certain cases was perhaps the most significant finding. Hopefully, this will lead to more frequent equipment checks, potentially utilizing our self-developed method.”
What have been the main challenges in your doctoral project?
”Ultrasonics technology research has been novel for both me and my supervisors, which made it more challenging. However, there’s been significant experience in observational studies, which has been a major advantage.”
Text: Jakob Lundberg
Cover illustration: The image to the left shows two electrical measurements of the sensitivity of the piezoelectric elements from a transducer that has several defective elements. To the right SDR (Systematic Dark Region) curves are shown for the same transducer for a long period. The occasions for the electrical measurements are marked with multi-colored lines. The SDR curves are blue where the transducer has defects.
Figure 7: A screenshot from the console used to manage the information from the new method where SDR (Systematic Dark Region) curves for several transducers are handled. Left: scanner/transducer. Middle: median image (top) and area under curve for the SDR curve (bottom). Right: Single SDR curve (top) and SDR curves 3D over time (bottom).