Deep learning and its applications have grown in recent years. Recently, researchers from ETH Zurich used the technique to study dark matter in an industry first. Now, a team working with the University of California, Berkeley and the University of California, San Francisco (UCSF) School of Medicine have trained a convolutional neural network dubbed "PatchFCN" that detects brain hemorrhages with remarkable accuracy.
In a paper titled "Expert-level detection of acute intracranial hemorrhage on head computed tomography using deep learning", the team claims that:
We used a single-stage, end-to-end, fully convolutional neural network to achieve accuracy levels comparable to that of highly trained radiologists, including both identification and localization of abnormalities that are missed by radiologists.
The team achieved an accuracy of 99 percent, which is the highest recorded accuracy to date for detecting brain hemorrhages. In some cases, the neural network's performance eclipsed even that of seasoned radiologists:
Our algorithm demonstrated the highest accuracy to date for this clinical application, with a receiver operating characteristic (ROC) area under the curve (AUC) of 0.991 ± 0.006 for identification of examinations positive for acute intracranial hemorrhage, and also exceeded the performance of 2 of 4 radiologists.
PatchFCN was trained on a dataset of more than 4,000 CT scans from UCSF-affiliated hospitals using Nvidia V100 Tensor Core GPUs and Amazon Web Services. The training and analysis were done in a novel way whereby the team divided up the CT scans into segments that were each subsequently analyzed by the model. The team then experimented with the segment size to achieve the best results to increase the model's accuracy.
Furthermore, according to the researchers, each picture can be analyzed within seconds by their trained model. After analysis, the model, in addition to passing a verdict on the existence of a brain hemorrhage, also provides a detailed tracing and measurement of each hemorrhage.
In the context of a hospital, this can be a vital asset. PatchFCN will not only improve throughput but will also relieve pressure off of radiologists’, thereby improving their efficiency and productivity, the team believes.
For more information and the specifics of the study, you can refer to the paper published here.