What does a pathologist do and what is digital pathology?
A pathologist analyses and predicts diseases like cancer based upon the microscopic analysis of very small tissue samples. Tiny slices of tissue are coloured and fixed upon glass slides. The pathologist looks through a microscope to analyse the structure and cell deformations to determine the disease. Since a few years it is possible to make digital images of those glass slides. The field of viewing, sharing and analysing those digital images on a computer screen is known as digital pathology.
How did we get into digital pathology?
Because we are intrigued by deep learning, we decided to see if we would be able to recognise and classify digital pathology images using AI and deep learning techniques.
With AI and Image Recognition techniques it is possible to build and train algorithms that classify images into groups. To train those algorithms you need data (digital images) and labels (abnormal/normal tissue). After training the algorithms, the algorithm can predict the label of a new, unseen image. The performance of the algorithm depends highly upon good quality data and correct labels.
The aim of the study was to build an algorithm that would predict the chance of a tissue being abnormal. So the digital images were classified into two groups: normal and abnormal tissue.
We built and trained several algorithms. Afterwards the different algorithms were tested on unseen images.
The best performing algorithm achieved an accuracy of 95% on unseen images, with 98% precision and 99% recall on abnormal tissue. The algorithm achieved 88% precision and 89% recall on normal tissue. Because the dataset was small, we are confident that we will be able to build algorithms that will perform with high accuracy, precision and recall.
With these promising results the IxorThink team is very excited to continue this journey. The final goal is to implement an end-to-end product which will help the pathologists to prioritise their workflow.