Figure 5.

We used the Grad-CAM model interpretation technique to obtain a class discriminative localization map for each prediction. We first computed the gradient of the class of interest (before the softmax function) with respect to feature maps of the last convolutional layer in the Resnet. These gradients flowing back are global average-pooled to obtain the neuron importance weights for the target class. A heat map of location importance is then up sampled to match the image size and overlaid on the input image. We then leveraged the invertible property of our spherical transformation method to generated articular surface importance heat maps for model interpretation for each bone and for each single biomarker. This process was performed on the first timepoint of every unique patient in the hold out test set (n = 875) and is illustrated for the femur.