Self-supervised vision-language pretraining for Medical visual question answering