no longer supports your web browser (Internet Explorer version 9 or lower). See what browsers we support.

Morphology Miasma: Using AI and ML to Improve Pattern Recognition Skills

The introduction of Artificial Intelligence (AI) and Machine Learning (ML) to the healthcare space presents a unique opportunity to promote learning and improve pattern recognition when making a skin diagnosis. This is especially relevant when determining morphology as it is vital to identify the lesion correctly to come to an accurate diagnosis for a skin concern.

Basic Morphology

Fifteen percent of all doctor visits involve a skin complaint and up to 48% of the time, patients are misdiagnosed1. If the lesion type is identified correctly, it is much more likely that a clinician will reach the right diagnosis. Here are a few basic lesion types:


flatmacule.pngFlat macule - A flat, generally less than 1.0 cm area of skin or mucous membrane with different color from surrounding skin.



flatpatch.pngFlat patch - A flat, generally greater than 1.0 cm area of skin or mucous membrane with different color from surrounding skin.


smoothpapule.pngPapule - A discrete, solid, elevated lesion usually less than 1.0 cm in diameter. Papules may be further classified by shape, size, color, and surface change.



smoothplaque.pngPlaque - A discrete, solid, elevated lesion usually broader than it is thick, measuring greater than 1.0 cm in diameter. Plaques may be further classified by shape, size, color, and surface change.



vesicle.pngVesicle - A fluid-filled cavity or elevation less than 1.0 cm in diameter. Fluid may be clear, serous, hemorrhagic, or purulent.

bullae.pngBulla - A fluid-filled blister greater than 1.0 cm in diameter. Fluid can be clear, serous, hemorrhagic, or purulent.

pustule.pngPustule - A circumscribed elevation that contains pus. Pustules are usually less than 1.0 cm in diameter.



Scale and Crust

scale.pngScale - Excess stratum corneum accumulated in flakes or plates. Scale usually has a white or gray color.

crust.pngCrust - A hardened layer that results when serum, blood, or purulent exudate dries on the skin surface. Crusts may be thin or thick and can have various colors.

eschar.pngEschar - An adherent thick, dry, black crust.


petechiae.pngPetechiae - Tiny 1-2 mm, nonblanchable, purpuric macules resulting from tiny hemorrhages.

ecchymosis.pngEcchymosis - Extravasation of blood into the skin or mucous membrane, forming large macules or patches.

palpable_purpura.pngPalpable Purpura - Raised and palpable, nonblanchable, red or violaceous discoloration of skin or mucous membrane due to vascular inflammation in the skin and extravasation of red blood cells.



ucler.pngUlcer - A circumscribed loss of the epidermis and at least upper dermis. Ulcers are further classified by their depth, border, shape, edge, and tissue at the base.

erosion.pngErosion - A localized loss of the epidermal or mucosal epithelium.



In addition to lesion types, there are also 10 basic lesion shapes and configurations: annular, reticular, arcuate, scattered, grouped, serpiginous, linear, targetoid, polycyclic, and whorled. LearnDerm is a free resource to better understand lesion types and basic dermatology.

These are the fundamental lesion types one might see during a skin exam. However, there are hundreds of morphology types and terms that exist. It would be extremely difficult for a nonspecialist to be armed with this information at the point of care based on memorization alone.

Misdiagnosis leads to higher costs and wasted time, and there are not enough dermatologists to handle the case load. This is where ML and AI can come in to help provide important information and improve the recognition of morphology at the point of care.


DermExpert was developed to assist clinicians at the point of care when they are faced with a skin condition. DermExpert allows users to take a picture of a lesion. It then uses machine learning to help narrow down the morphology (trains on 86 types) as well as the patterns that are associated with certain diseases and lesion types. Getting a clinician to a meaningful differential is critical to reaching the correct diagnosis.

The software also aims to train clinicians in the language of the skin exam while providing guidance and helping to build a meaningful differential. By improving pattern recognition and providing extra support, DermExpert can help non-dermatologists become more confident in diagnosing skin conditions.

DermExpert is now available for iOS and Android devices. Learn more and schedule a demo on our website.


1. Fleischer AB. Diagnosis of Skin Disease by Nondermatologists. AJMC. Published October 1, 2000.

About VisualDx

VisualDx is an award-winning diagnostic clinical decision support system that has become the standard electronic resource at more than half of U.S. medical schools and more than 1,500 hospitals and institutions nationwide.  VisualDx combines clinical search with the world's best medical image library, plus medical knowledge from experts to help with diagnosis, treatment, self-education, and patient communication. Expanding to provide diagnostic decision support across General Medicine, the new VisualDx brings increased speed and accuracy to the art of diagnosis. Learn more at