Deep learning in label-free cell classification and machine vision extraction of particles

A computer-based automated method of sorting particles for flow cytometry, the method comprising:

• training, by a training device, an artificial neural network (ANN) which receives inputs as digitally observed features of known particles that belong to at least two known particle classes;
• tuning, by the training device, weights of the ANN as a function of at least one property of receiver operating characteristics (ROC) curve representing particle classifications of the known particles generated by the ANN based on the digitally observed features;
• configuring a sorting device with the trained and tuned ANN;
• storing in a computer readable non-transitory memory, by the sorting device, particle sensor data relating to observed particles that are observed by a sensor;
• converting, by the sorting device, the particle sensor data into digitally observed particle features representing the observed particles;
• and
• sorting, by the sorting device, the observed particles into the at least two classes according to the trained and tuned ANN as a function of the observed particle features;
• wherein said method is performed by executing instructions on a computer processor, and wherein said instructions are stored on a non-transitory memory readable by said computer processor.