CALM (Catch And Light-up Marker)

Abstract

Aim

This study aims to assist radiographers in accurately placing anatomical markers during radiographic procedures by leveraging computer vision and AI technologies. The system seeks to reduce human error and enhance workflow efficiency, which is essential for ensuring precise diagnosis and treatment.

Methods

CALM integrates AI-driven models with automated monitoring tools, including scanners and cameras, to capture data from laboratory instruments and visual observations. It uses real-time screen monitoring via an HDMI splitter to detect X-ray orders, extract “R” or “L”, and display the corresponding marker on an LED screen next to the X-ray machine.

Results

In the initial 10 tests, when X-ray orders were captured from the screen, CALM successfully identified the right and left sides and correctly displayed “R” or “L” on the LED screen with 100% accuracy. However, when the orders were captured from paper forms, the accuracy dropped to 90%, likely due to OCR limitations in extracting the correct side information from the printed text.

Conclusion

In conclusion, CALM has proven to be useful in accurately identifying and displaying the correct side (right or left) in X-ray orders captured from the screen, achieving 100% accuracy. However, this study is limited to orders that specifically contain the words “right” and “left.” For orders with other terminology, different approaches will need to be further investigated and developed to ensure consistent and accurate marker placement.