|
Statistical Medical Image Database MDOL, Inc | ||
| Structured Medical Image Database | ||
MDOL is building a large, well-organized database of medical
images and associated clinical reports. The database currently
consists of over 20,000 patients and over 3,000,000 medical images and
provides a rich source of information for research or clinical use.
The data were acquired from over 15 top medical institutions around
the world. All the information was anonomized at the institution, and
our database is 100% HIPPA compliant.
The plot to the left shows just one example of the types of statistics we can derive from the database. Here, we correlate occurrence of disease as a function of age. Each curve (representing a disease) is area normalized. There are many other types of relationships that can be extracted and correlated from the information contained in our database. |
||
| Image-Based Pathology Statistics | ||
Cases in our database include over 150 pathologies of the brain.
The pathology is localized in each abnormal image. Organizing the
image database in this way provides useful statistics about anatomy
and pathology.
The image to the right shows a spatial probability distribution of pathology derived automatically from our database. Red indicates where each pathology is most likely to occur, followed by yellow, green, and blue. Where there is no color, our database has not seen an example of that patholoogy in that location. While both multiple sclerosis and small vess ischemia are white matter diseases, notice the subtle differences between their spatial probability maps as derived from our database. Other types of pathology-based information that exists in our database include lesion size, location, quantity, changes over time, etc. These factors can easily be correlated with other factors, just as age, gender, history, treatment, etc. |