Design of A Novel Device Measuring Testicular Consistency, Size and Temperature and Classification Algorithm Analysis of Data Compared to Urologist Examinations

Öz

Background and Objective: Infertility is a health problem that can be seen all over the world, originating from both men and women. In male infertility, many tests and examinations are performed in order to find the treatment method to solve the problem. The first of these is the physical examination, and the clinician cannot obtain numerical data and make comparisons as a result of this examination.

Methods: In this study, it is aimed to design a system that will reveal the physical differences of the testicles, which play an important role in the production of sperm and some special hormones, in about 1 minute. Thanks to this device, temperature, volume and consistency parameters of testicles from 50 different patients of different ages were measured in real time with different sensors on the device. The results were analysed with both statistical analysis and machine learning method.

Results: According to the results of this study, it has been revealed that the system designed can help clinicians in testicular examination. In terms of consistency and heat stress, the classification algorithm with the highest accuracy rate according to the 3 different cross validation rates applied is trees and it showed 92.0% accuracy according to validation 5 and 15, but 89.0% according to cross validation 10. In terms of testicle size, Trees and Ensemble with the validation rate of 15 showed the highest accuracy with 89%.

Conclusions: Although the applied methods showed high accuracy, the specificity rate is not optimal due to data limitations. This is because there is not enough data. It is clear that with more data, both the accuracy rates and the level of specificity will be higher.