Evaluating Image Training Systems for Medical Students

Abstract

Skin cancer is one of the common and most fatal cancers. In most cases, the similarity between benign (healthy) and malignant (harmful) makes it so difficult to diagnose the lesion correctly. Moreover, there are two levels of categorization for skin lesions. In addition to benign vs malignant (basic level), each skin lesion can also be categorized as one of the sub-types of benign or malignant (subordinate level). In most medical schools the distinction between skin lesions is taught to students in just four sessions and at the basic level - i.e. benign vs malignant.

In this research, we designed a learning system which can assist students in learning skin lesions effectively in only a few sessions through an application using skin lesion images. We also compared these two levels, basic level and subordinate level, and found that indeed learning skin lesions at the basic level is more effective at distinguishing harmful cases than at the subordinate level as it could be hypothesized.