CT workflows improve with help of deep learning 3D camera for patient positioning
Use of a 3D deep learning camera on CT gantries could speed workflows by optimizing patient positioning.
The rising demand for CT imaging has prompted numerous efforts to improve workflows without encroaching on image quality or patient care. Various forms of artificial intelligence have been developed to address common workflow hiccups, such as contrast dosing, image reconstructions and patient positioning. Deep learning 3D cameras, which automate patient positioning based on their habitus and placement on the table, also have garnered interest in recent years as a simple way to shorten exam times.
“A deep-learning 3D camera can automatically detect the patients’ body surface contour and their positions. Previous studies have reported that patient positioning by a deep-learning 3D camera can reduce positioning time compared with positioning by radiographers,” Yoshifumi Noda, MD, PhD, with the department of radiology at Mass General Brigham, and colleagues explained in the European Journal of Radiology. “Moreover, previous studies have reported that the patient positioning by a deep-learning 3D camera allows to reduce the deviation between the scanner and patients’ isocenters as well as reduce the radiation dose compared with manual positioning by radiographers.”
Recently, the group sought to determine exactly how use of deep learning-based 3D positioning assistance affected workflows and image quality. To do this, they compared the unenhanced chest-abdomen-pelvis CT scans of a group of nearly 600 patients between October 2023 and January 2024; patients were divided into either a manual positioning group or a 3D camera group, with radiologists comparing how each method affected image quality, exam duration and patients’ total radiation dose.
Both positioning methods yielded similar measures of CT dose-index volume (CTDIvol), dose-length product (DLP) and background noise. However, use of the camera reduced the total room time from 255 seconds to 223 seconds. Meanwhile, median positioning time dipped from 79 seconds to 57 seconds, and scanning time from 86 seconds to 78 seconds.
Although this may not seem like a significant decrease, when compounded throughout the day with multiple scans, it does represent a marked improvement, the group suggested.
“We prospectively revealed that the use of a deep-learning 3D camera could improve the workflow of CT examinations compared with manual positioning by radiographers,” the authors concluded. “Although its impacts on the accuracy of patient positioning, radiation dose, and background noise were minimal, the deep-learning 3D camera can facilitate the workflow of radiographers in CT examinations.”
Read more about the findings here.
In addition to her background in journalism, Hannah also has patient-facing experience in clinical settings, having spent more than 12 years working as a registered rad tech. She began covering the medical imaging industry for Innovate Healthcare in 2021.