Automatic organ validation of b-mode ultrasound images for transmission to cloud

Miniaturization in size of Medical ultrasound scanning machine made it to use in point of care applications. Lack of sonographers and their unwillingness to work in rural areas limit the benefits of ultrasound system in rural healthcare. Diagnosis of patients through ultrasound is done by visualizing the ultrasound scanned images of organs. Diagnosis through telemedicine involves transmitting of ultrasound images from rural locations to cloud, where sonographer can remotely access the ultrasound data from cloud and generate the report, thus reducing the geographical separation between patients and doctors. Due to lack of adequate sonographers, ultrasound scanning in remote areas is operated by semi-skilled clinicians. Most of the images generated by semi-skilled clinicians are not useful for diagnosis. Transmitting all these images increases the data in cloud, drains the battery of portable ultrasound machine and increases latency in medication. This paper provides automatic B-mode ultrasound image validation based on organ information present in the image for diagnosis, thus avoiding transmission of invalid images to cloud. Linear kernel SVM classifier trained with first order statistic features of image with/without organs is used to classify the images into valid and invalid for diagnosis. The algorithm resulted with a recognition efficiency of 94.2% in classifying the ultrasound images.