What the post claims, and what it leaves out
The post makes three separate claims:
- Some CT scanners use the Kinect.
- The 3D body map is used for diagnosis.
- The Kinect was a cheap alternative to professional 3D sensors.
The post names no manufacturer, scanner model or study, so the first claim can't be traced. The second is wrong in kind. Per Pasquale Pillitteri's write-up, a camera above the couch diagnoses nothing. It helps position the patient, choose where the scan starts and estimate body size so the dose can be tuned. The third is the most defensible, because a console accessory did put depth sensing within reach of university labs.
The Kinect in CT research
The best-known Kinect-to-CT study came from the Hospital of the University of Pennsylvania. Its Journal of Digital Imaging paper, with Tessa Cook as lead author, proposed estimating patient volume with the Microsoft Kinect, a combined RGB camera and infrared depth sensor. The motivation was that height, weight, BMI and effective diameter are one- or two-dimensional measures that don't capture the size of the anatomy being imaged.
According to Vision Systems, an algorithm from Radimetrics produced a depth map and skeletal outline of each person, which were used to estimate whole-body volume. Posture was a problem. With arms raised, bodies appeared larger and density estimates fell. With arms crossed, they appeared smaller and estimates rose. Cook suggested two Kinects at 90 degrees, or sequential frontal and lateral depth maps from one camera, might be needed.
Pillitteri's report gives the numbers. The test used five adults, and the density estimates were 995.2 kg/m³ with arms at the sides, 805.1 with arms raised and 1,136.8 with arms crossed. Water is about 1,000 kg/m³, so the swing was large. Medgadget's coverage from the time confirms the setup. The person stood 2.5 meters from the camera for several seconds while a depth map and skeleton were computed. This was a proof of concept, not a scanner feature.
A closer link: a Kinect wired to a CT scanner
One later study comes closer to the viral claim than the Penn work does. A 2018 BioMed Research International paper reports that researchers integrated a somatosensory controller with a CT scanner to measure patient anterior-posterior and lateral diameters and calculate effective diameter. It enrolled 108 individuals and used a Microsoft Kinect to acquire depth images.
The mean absolute differences were modest. Against an F ruler they were 5.2% (APD), 1.3% (LATD) and 2.5% (ED). Against CT images they were 8.8%, 2.6% and 5.0%. The authors concluded that the approach was simple, low cost, free of radiation and automatic.
This matters for the headline question. A Kinect was integrated with a CT scanner in a research setting. That is not the same as a commercial scanner shipping with a Kinect. Nothing in the material reviewed here shows it as a standard component.
What the cameras on modern scanners are
The Erasmus MC studies in European Radiology describe scanner-integrated 3D cameras. None of them identifies the device as a Kinect.
- 2018, adults. The study compared 423 patients positioned manually with 254 positioned using a camera's table-height suggestion. The camera used infrared body-contour detection and fitted a scalable patient model. Median deviations from ideal table height were 12.0 to 14.7 mm for radiographers and 5.4 to 9.5 mm for the camera. The authors stress that radiographers remain indispensable when the system fails or the case is difficult.
- 2020, children. The scanner was a Siemens SOMATOM Drive equipped with a commercially available Siemens 3D camera. The camera's color and depth images feed an algorithm that fits a virtual patient avatar. In children without fixation aids, the camera's median deviation was 3.7 mm against 10.3 mm for radiographers. With baby cradles or vacuum cushions the avatar usually couldn't be fitted, and the difference was not significant. The improved algorithm was described as not commercially available.
- 2023, preprint. Siemens-affiliated researchers proposed estimating scan range, isocenter and Water Equivalent Diameter from 3D camera images, to remove scout scans from lung cancer screening. The model was trained on over 60,000 CT scans and tested on 110 depth-and-CT pairs. Average errors were 5 mm for isocenter and 13 mm for scan range, and relative WED error was 4% against a 10% IEC criterion. It is a preprint, and it says "3D camera," not Kinect.
Why does positioning matter? The pediatric paper explains that off-center positioning affects the patient's apparent size on the localizer radiograph and therefore how automatic exposure control behaves. Better centering means more consistent dose and image quality. That is a practical benefit, but a modest and indirect one.
Does it save lives?
No study here measures mortality. They measure millimeters of table-height error, density estimates and percentage differences in body diameter. The Siemens preprint mentions lung cancer and early detection only as background for automating the workflow. Anyone saying the Kinect "saved lives" is going beyond the data.
Caveats
- Most of this evidence is single-center. The Erasmus MC work is retrospective, and the Siemens work is a manufacturer-affiliated preprint.
- The Kinect studies were small or exploratory, and the five-volunteer Penn experiment was an early prototype.
- Absence from the published literature doesn't prove no hospital ever used a Kinect locally. It only means no study documents it as the camera in a commercial scanner.
- The Pillitteri write-up reports Microsoft ended Kinect production in 2017 and the Azure Kinect Developer Kit in 2023. That is another reason a hospital buying new equipment would not choose it.
Bottom line
The accurate version is that the Kinect showed up in CT research, including a 108-person study that integrated it with a scanner. Commercial scanners that use depth cameras carry the manufacturers' own. The Kinect's real contribution was lowering the cost of experimenting, and that is a market fact more than a medical one.