“Artificial intelligence will digitally disrupt all industries. Don’t be left behind.” -Dave Waters
It is no secret that in recent years artificial intelligence/AI (along with machine learning and deep learning) have exploded in the healthcare industry. Only a few examples of this include utilizing advanced big data processing algorithms for predictive analytics, business intelligence, and personalized care. Perhaps one of the areas affected most by this revolution is the dermatoscopy industry, due to the fact that dermatology largely regards the examination of the skin.
The Current Status Of Dermatology Imaging
For example, when a patient arrives for an appointment, they will likely have clinical or dermoscopic images captured of their skin for followup reference. Similarly, teledermoscopy software is allowing for dermatologists to receive high quality dermatoscopy images directly from their patients.
When searching for this subsequent dermatoscopic image, you typically choose a few keywords that describe it and then submit them into a search engine. The problem that arises with this method is that many images are simply too complex to describe, requiring a detailed description to correctly capture the shape, size, texture and color.
Unless these images are properly labelled and organized, they can be very challenging to find for subsequent followup appointments, or for reference as an educational resource for potentially similar cases with future patients. The question becomes how can artificial intelligence support this process?
AI: Evolving The Way You Practice Dermoscopy
Rather than trying to come up with keywords accurately depict your desired image, it is much easier use Content-Based Image Retrieval (CBIR). Using this technique, you no longer have to provide written information; instead, the visual features within an image are used to visually retrieve similar pictures.
Intelligent dermatology software such as DermEngine's Visual Search provides an easy solution for medical professionals to find similar cases. A key aspect that powers Visual Search is the use of a convolutional neural network (CNN). CNNs take an image and convert it into a compact vector representation (a series of numbers that the computer can easily process) whilst retaining the visual information found in the original image. In turn, these vectors are utilized to efficiently retrieve dermoscopy images that appear to be visually similar.
Applications and Benefits
In this image returned from DermEngine's intelligent dermatology software, the moles are visually similar to each other. They are mostly round and dark spots that the system has identified as being similar, while understanding that the skin color is not an important factor when deciding which images to retrieve.
Intelligent dermatology software such as Visual Search drastically transforms the way medical professionals can perform consultations. When providing feedback to a patient case, dermatologists are now equipped with intelligent tools designed to support their clinical decisions for streamlined care and the early detection of skin cancer and other skin conditions for improved patient outcomes.
Additionally, medical professionals can use intelligent clinical decision support tools like these as an educational resource. Experts can utilize the database of pathology-labelled images obtained from dermatologists around the world to help train or teach medical students, researchers or non-expert physicians.
Artificial intelligence (AI) is currently tapping into virtually limitless potential in the field of dermoscopy. Content-based image retrieval (CBIR) algorithms allow professionals to quickly locate images, receive support in their clinical decisions, and utilize the retrieved images as an educational resource. Perhaps most importantly of all, due to the fact that these cloud software use machine learning algorithms, its accuracy is constantly improving. As time continues to pass, it is clear that CBIR tools will be at the forefront of supporting dermatologists so that they can provide enhanced care to their patients.
-The MetaOptima Team