Hi Mark,
Based on the images you sent, I entered the features into the Stanford Bones Jones Bayes Classifier.
Chris and I are still working on the model and it is highly biased to the data we used (Dahlin, an old collection ~700 tumors by Dr. Jones at Stanford, and 40+ publications), so for what it’s worth, it says consider chondroblastoma or ganglion/cyst. The advantage of Bayes theorem is explainability using intuitive odds which are used to calculate the posterior probabilities that determine the order of the differential diagnosis… In the graph below, a near horizontal (not steep) step means it’s a strong factor, and in the table below, the higher the factor, the more likely that observed feature contributes to the higher posterior probability. Not shown, but low factors are helpful to exclude disease, ie, "This is not GCT b/c only 1/100 have a sclerotic transition zone" –> made that up)… The ddx generated are useful as suggestions which can then be correlated w/ clinical experience and Google…
This new Bayes model is not yet validated/published, so pls take it with a grain of salt !
