I implemented an image segmentation pipeline, train on the [DICOM dataset][1]. I compare the results of the model with manual segmentation to find the accuracy. Is there other methods for evaluation? [1]: https://en.wikipedia.org/wiki/DICOM