ALIAS
Automatic Localisation and Identification of Anatomical Structures in Medical Images (ALIAS)
Due to the rising number of images and their improved resolution in medical applications, there is a high need for fully automatic algorithms in medical image processing, which are not in need of user interaction.
In the scope of the project ALIAS we therefore want to develop an algorithm for the automatic and robust localization of anatomical structures in medical images. The information of object position is useful, e.g., for the automatic segmentation of organs or bones or the automatic measurement of length and angles. The developed procedure should thereby be easily applicable to new target organs or modalities (x-ray, CT, MRI).
To deal with this task, we use the Generalized Hough Transform (GHT) in combination with a discriminative training algorithm (DMC) to train a weighted model. This model should consist of a small number of points representing our target object, which is capable of localizing our target in unknown images. The approach has been testes on several tasks (Publications), like the localization of joints in long-leg radiographs or the liver in abdominal CT-images.
However, there is still a large number of possible improvements and questions we have not yet dealed with. Students, who are interested to work with us in the project, e.g., in the scope of a thesis or master project, are welcome to contact Prof. Hauke Schramm or Heike Ruppertshofen.
- Project leader: Hauke Schramm
- Research Associates
- Heike Ruppertshofen
- Ferdinand Hahmann
- Students
- Gordon Böer
- Ralf Stannarius
- Inga Berger
- Former Students
- Francesco Boero
- Markus Brunk
- Daniel Großmann
- Daniel Künne
- Junaid Naseer
- Jan Schlichting
- Jun Zhang