MRI Angle Measurement Inter-rater Reliability

I am a physician working on a self-funded study. To give you some background, we are working on a safer technique for one of the spinal injection procedures we perform. We have already published two papers on this (attached), one validating the technique and another showing safety data. The one we are currently working on is an inter-rater reliability study as our new technique requires an angle measurement on an MRI. This is retrospective and IRB approved (exemption) obtained from Stirling IRB. Three physicians (myself, another experienced physician and a trainee) have reviewed 50 different MRI's and measured a patient specific angle independently (data attached). We need to do inter-rater reliability stats on this data. To clarify, each of the 50 patients has unique angle that was measured independently by each of the 3 physicians. Figured Kappa or anything else you would suggest. It is a bit unusual in that we need to compare the raters all to each other not taken all together (rater 1 compared to rater 2 and rater 1 compared to rater 3 and rater 2 compared to rater 3). The reason for this is that rater 1 and 2 are experienced physicians and rater number 3 is a trainee. Our field also seems to crave categorical data. We were thinking that any difference between raters of 5 degrees or less (five degrees does not make much of a clinical difference) is fantastic so coming up with categorical data as well would be useful. For example "X% of the measurements between rater 1 and 2 were within 5 degrees" You will see on subject 15, there is one measurement that had to be thrown out because rater number 3 (the trainee) measured the wrong side. We kept the data on that subject for rater 1 and 2 for comparison.