3. Conclusions

As mentioned before, conventionally, BCD and classification is performed by a clinician or a pathologist by observing stained biopsy images under the microscope. However, this is time consuming and can lead to erroneous results. Therefore, there is a rise in the need for developing intelligent and automated technology. As this problem can be computationally modeled as a problem of retrieving relevant information from a plethora of reports or tests, the development of a computationally efficient and detection-wise effective system for BCD can be seen as an issue from the field of KDD. Being cross-dimensional, KDD uses algorithms and techniques from a vast array of fields like soft computing, pattern recognition, machine learning statistics, artificial intelligence (AI), natural language processing (NLP), etc.

In this research, a method based on advanced DIP, KDD, and AI techniques was used for the extraction of fundamental features in DIM in order to detect breast lesions by using a GRANN. The used methodology was divided in two main stages. The first one used advanced DIP techniques for extracting image features of DMI in order to create a biomarker for BCD. With this information, in the second stage, a GRANN was trained and tested in order to classify BC.

After 2000 network trainings, a smoothing factor equal to 1e4 was calculated. This value was used for training the neural net reaching an accuracy of 95.83%. The performance of trained GRANN was evaluated using four performance measures: accuracy, sensitivity, specificity, and precision. The results obtained in this work show that GRANN is a promising and robust system for BCD.

In a third stage, a CADx system based on AI technology is being designed in order to be applied in Mexican patients in collaboration with GHZ and MML. The proposed system aims to eliminate the unnecessary waiting time as well as reducing human and technical errors in diagnosing BC. The Laboratorio de Innovacion y Desarrollo Tecnologico en Inteligencia Artificial is seeking collaboration with research groups interested in validating the technology being developed.
