Deep learning models for region of interest determination

An image processing system comprising:

• at least one computer readable non-transitory memory storing software instructions, a first description of a structure type of a first region type, and a second description of a structure type of a second region type that is different than the first region type;
• and
• at least one processing circuitry coupled with the memory and that performs, upon execution of the software instructions, operations of:
• partitioning a digital image into a set of areas;
• identifying a structure type of each area of the digital image;
• determining regions of interest in the digital image by applying a classifier able to determine regions of interest based on structure types of the set of areas in the digital image, including determining a first region of interest in the image by matching the set of areas of the image with the first description of the first region type, and determining a second region of interest in the image by matching the set of areas of the image with the second description of the second region type;
• and
• presenting the regions of interest with different visual appearances indicating respective region types.