**5. Conclusions**

In this work, we proposed a new weighted voting ensemble self-labeled algorithm for the detection of lung abnormalities from X-rays, entitled WvEnSL. The proposed algorithm combines the individual predictions of self-labeled algorithms utilizing a new weighted voting methodology. The significant advantage of WvEnSL is that weights assigned on each component classifier of the ensemble are based on its accuracy on each class of the dataset.

For testing purposes, the algorithm was extensively evaluated using the chest X-rays (Pneumonia) dataset, the Shenzhen lung mask (Tuberculosis) dataset and the CT Medical images dataset. Our numerical experiments indicated better classification accuracy of the WvEnSL and demonstrated the efficiency of the new weighted voting scheme, as statistically confirmed by the Friedman Aligned Ranks nonparametric test as well as the Finner post hoc test. Therefore, we can conclude that the new weighted voting strategy had a significant impact on the performance of all ensembles of self-labeled algorithms, exploiting the individual predictions of each component classifier more efficiently than the simple voting schemes. Finally, it is worth mentioning that efficient and powerful classification models could be developed by the adaptation of ensemble methodologies in the SSL framework.

In our future work, we intend to pursue extensive empirical experiments to compare the proposed WvEnSL with other algorithms belonging to different SSL classes, and evaluate its performance using various component self-labeled algorithms and base learners. Furthermore, since our preliminary numerical experiments are quite encouraging, our next step is to explore the performance of the proposed algorithm on imbalanced datasets [39,40] and incorporate our proposed methodology for multi-target problems [41–43]. Additionally, another interesting aspect is the use of other component classifiers in the ensemble and enhance our proposed framework with more sophisticated and theoretically sound criteria for the development of an advanced weighted voting strategy. Finally, we intend to investigate and evaluate different strategies for the selection of the evaluation set.

**Author Contributions:** I.E.L., A.K., V.T. and P.P. conceived of the idea, designed and performed the experiments, analyzed the results, drafted the initial manuscript and revised the final manuscript.

**Funding:** This research received no external funding.

**Conflicts of Interest:** The authors declare no conflict of interest.
