Author(s): Ramez Barsoum, Resident Medical Officer, Princess Alexandra Hospital, Brisbane, Australia. DermNet New Zealand Editor in Chief: Adjunct Prof Amanda Oakley, Dermatologist, Hamilton, New Zealand. Copy edited by Gus Mitchell. September 2018.
A convolutional neural network (CNN) is a type of deep artificial neural network used in image processing . It is a multifaceted means of analysing data that involves complex mathematics and requires massive computational power to combine biology, mathematics and computer science. It draws on similarities from nature, where neurons are interconnected, capable of learning and improving on performance, based on how many lesions it ‘sees’.
The visual nature of dermatology lends itself well to digital lesion imaging and CNNs have huge potential to change practice. Diagnosing lesions by a dermatologist generally involves an input image (of a cutaneous lesion) being fed through a processing network (the skills and knowledge of the dermatologist who analyses it and synthesises available information) to output a ‘class’ (or diagnosis) or a ‘probability of classes’ (differential diagnosis) .
CNNs have been used in military and civil applications, in the technology sector, and in commerce. They are found in everyday applications such as social media platforms that automatically recognise faces, in self-tagging photo galleries, and on shopping websites that come up with suggestions based on your browsing habits.
In the medical field, researchers have been using CNNs to diagnose diabetic eye disease, arrhythmias, and skin cancers [2–4].
The basis of a CNN is a computer program that is able to differentiate between different image classes based on unique features that can reliably be used to identify the image class, such as edges and curves. This is expanded upon with more abstract features through a series of convolutional layers.
An image is fed into a computer and processed as different arrangements of dots (pixels), based on their colour. This converts sections of the image according to a particular filter . The filters often start off as simple features like straight lines, diagonal lines, curved lines, or dots. Every time a filter is passed over the original image, it creates a new, smaller version of the original photograph. The positive filter matches are assigned a positive value, and the areas that do not match are assigned a value less than 1. This results in a convoluted image: for example, a straight-line filter passed over an image of an acral naevus with parallel furrow pattern on dermoscopy will show a strong positive convolution image. This step can be repeated for more features to achieve a more accurate output.
If the resulting image is large, it may be required to “pool” layers between subsequent convolutions . There are different types of pooling, but the most common type is the maximum pooling method (figure 1).
There is often an additional normalisation step.
To generate an output differential diagnosis for a suspected lesion, the neural network needs to apply a fully connected layer based on all the layers it has previously processed. This is akin to a dermatologist synthesising the different clues into a provisional diagnosis with a set of differentials.
The CNN can now be trained using additional functions to improve accuracy and to “teach” itself to identify new lesions (such as back-propagation, which teaches the network when it selects a wrong outcome to change the weight assigned to traits when selecting an output class) .
Benefits of CNNs in the diagnosis of skin lesions include accuracy, speed and low cost.
Caveats include unrealistic expectations of patients and health practitioners, security and privacy issues, and medicolegal accountability.
See the DermNet NZ bookstore.
© 2018 DermNet New Zealand Trust.
DermNet NZ does not provide an online consultation service. If you have any concerns with your skin or its treatment, see a dermatologist for advice.