Computer input devices: neutral party or source of significant error in manual lesion segmentation?

Area of Science:

• Medical Imaging Analysis
• Radiology
• Image Segmentation

Background:

• Lesion segmentation is critical for tracking diseases in medical imaging.
• Accurate boundary delineation distinguishes normal from abnormal tissue.
• Input devices impact segmentation efficiency and accuracy.

Purpose of the Study:

• To compare the accuracy and efficiency of graphics tablets versus laser optical mice for lesion segmentation.
• To evaluate radiologist preference and subjective experience with different input devices.

Main Methods:

• Eleven radiologists segmented 132 simulated lesions using both a graphics tablet and a laser optical mouse.
• Contour and area segmentation errors were quantified using root mean square error and comparison to reference lesions.
• Segmentation time and subjective user experience were recorded.

Main Results:

• Graphics tablets significantly reduced contour segmentation error compared to mice (P < 0.0001).
• Area segmentation error showed no significant difference between the two devices (P = 0.62).
• Segmentation time was significantly shorter with the graphics tablet (P = 0.011), and users overwhelmingly preferred it.

Conclusions:

• Graphics tablets offer superior accuracy for lesion contour segmentation, with a slight speed advantage.
• For lesion area segmentation, both graphics tablets and mice provide comparable accuracy.
• User preference strongly favors graphics tablets due to perceived speed, ease of use, and accuracy.