In the past section, ROC analysis has been analysed. Instead of a pixel-based evaluation, Free Response Operating Characteristic (FROC) is based on a region-based analysis [26,120]. The FROC paradigm is, nowadays, being increasingly used in the assessment of medical imaging systems, particularly in the evaluation and comparison of CAD algorithms [16,82].
FROC analysis is similar to ROC analysis, except that the false
positive rate on the 
-axis is replaced by the number of false
positives per image. Thus, FROC seeks location information from
the observer (the algorithm), rewarding it when the reported
disease is marked in the appropriate location and penalizing it
when it is not. Note that this task is more relevant to the
clinical practice of radiology, where it is not only important to
identify disease, but also to offer further guidance regarding
other characteristics (such as location) of the disease.
Before FROC data can be analyzed, a definition of a detected
region is needed. Although there are different opinions in the
literature [46,82,133], in our work we
use a typical approach which expects a 
overlap between the
annotated and detected regions to indicate a true positive.