Optimisation of User Interaction for the Calculation of the Cobb Angle in Lumbar Spine MRI Images

Clinical studies conducted on large populations of patients require computer-based tools to efficiently and accurately process and interpret large amounts of imaging data. However, certain tasks are still performed manually. As a result, when a manual task needs to be repeated across a cohort of hundreds of patients, it can become a bottleneck in the analysis workflow. One example of these manual tasks is the computation of the Cobb angle. To compute the Cobb angle, users typically identify four points on spine images: two points for the tangent line to the upper end-plate of the upper vertebra and two points for the tangent line to the lower end-plate of the lower vertebra. This process requires four mouse clicks (a 4-point definition) for user interaction with the graphical interface. This article introduces a method where users need to specify only two points on the image for Cobb angle calculation, significantly streamlining the process. The main stages include: selecting vertebrae using two mouse click; extracting sub-images, segmenting vertebrae, and calculating bounding boxes; and finally, calculating tangent lines and determining the Cobb angle. Such efficiency is vital when managing images from a large patient cohort in a clinical study. The computational method has been published in an extension module for 3D Slicer. The evaluation of the method was performed on magnetic resonance images of the lumbar spine. The proposed method reduced the required interaction time by about 50% while maintaining an accuracy of 90%.