Artificial Neural Networks have emerged as the one of the leading algorithmic approaches to solve modern healthcare challenges such as biomarker detection, disease diagnosis and medical imaging.

The relatively large success of Artificial Neural Networks within this field can be attributed to the neural networks ability to identify underlying trends and abstract relevant details from a given dataset.

This paper presents a comprehensive review of Artificial Neural Networks by identifying the key working components of any Artificial Neuron starting from an external input to its corresponding output received from an activation function, it analyses the different methodologies to network construction such as Supervised Learning, Unsupervised Learning and Reinforcement Learning, while also comparatively reviewing popular network topologies used within neural network based healthcare solutions such as Convolutional Neural Networks and Auto Encoders.

Finally, this paper summarizes the current state of Artificial Neural Networks within Medico-Diagnostics, predicts future trends, and identifies upcoming areas of challenges such as overtraining of small datasets, a difficulty in obtaining large medical datasets owing to the amount of time it takes medical professionals to label data and a general lack in the availability of high-quality data.